#career-advice
1 messages · Page 291 of 1
hm yesh
8 hours of good sleep is better for health
r u in college or nah?
Damn golden advice ngl
No I am going to join the yr
And while we're at it https://www.youtube.com/watch?v=sTJ7AzBIJoI
THIS IS THE ORIGINAL MUSIC VIDEO
Great Music video from the nineties !
The lyrics are taken from a famous essay — written in 1997 by Mary Schmich, a columnist with the Chicago Tribune — which gives some amazing advice for life, thoroughly recommend everyone to watch this ! enjoy !
http://tinyurl.com/schmich-sunscreen
This is the original tun...
Such a helpful server
Hi
But yah, for begining programming advice: go to #python-discussion and ask there. Better channel for that question
@fringe sphinx done
Hello, I am looking for internships and possibly low paying junior roles, does anyone have recommendations for improving my CV or advice over the roles I should target?
Your CV currently doesn't say anything about the kinds of roles you're targeting, there's no quantification or indication of outcomes (what size of budget were you managing for the club, for example?), roughly half of it focuses on team projects. It's fine to have done things in a team, but be clear about what you have achieved in those teams. 'Non-trivial' is fluff without conveying anything meaningful to the reader.
Thank you so much for the review, I will implement what you said
because you focus in on this stuff so much and act so condescendingly towards others that it's kinda weird
Good luck landing a role!
dont interview with space x, that was actually miserable 😂
wasted like 20 hours of my time just to be interviewed by a guy who has no idea how to interview
Yes absolutely. I’ve generally found there are less jobs for low level stuff, but also less qualified applicants. Also quite a few jobs that want both and the pay for those are often quite high
yeah i made a project on computer applications and model training in this i have to train a model for factory workers wearing safety gears or not checking via CCTV and reporting a log of violation with the person
Is it okay if I fuh around in my high school years and learn programming in college?
why not learn in high school? programming can be really fun
said by someone who doesn't even study something programmin-related, im just a hobbyist
Idk learning from a teacher is better than learning by yourself
Not necessarily, maybe that's your way of learning, but I'm doing pretty good on my own. What have you tried so far?
Well I’m think I’m back to square one
So did you use OpenCV?
Learn or don't learn. You'll be fine either way.
I suck at learning programming by ms, I dose off during cs50, I can’t focus in books and other videos make me stuck in tutorial hell, I’m not “autonome” at all
We've given you advice for 2 years. It's always the same: start writing code.
Unc stop mentioning the 2 years thing I’ve changed since it hurts me when you mention my old self
I’ve heard of OpenCV but I don’t know much about it.
If you want to become a programmer, stop talking about it and write a simple program. Then write another one. And keep going. And soon enough, you'll be good.
And I haven’t tried to learn python for two years it’s been on and off
You don't need to follow a course. It helps. But it's not required.
Why don't you try to make something?
It's better to just start writting and investigating how to solve the problems you encounter.
I’ll follow bro code and after every tut I’ll make something and master the subject
Wow, 2 years? Then I should not be giving this guy advice, I only have like two months and a half, almost 3.
Dw I can explain in ot
What does this have to do with this server or channel?
!compban 1367043837108682763
:incoming_envelope: :ok_hand: applied ban to @indigo viper until <t:1779493060:f> (4 days).
Yo drop the 2 years thing pls
They haven't mentioned anything about that. Are you okay?
Tf is he saying @fringe sphinx
Do you find being rude and intentionally misgendering people to be effective?
??? Not my intention
Question: How can I go from "my code does what I want it to do" to "my code works efficiently"
If you don't have a reason to believe that your code isn't efficient, the default assumption is that it's efficient enough.
@jaunty storm what kind of code do you usually write? Different types of programs have different considerations
Are you afraid that if you post your successful cold reach out strategy here, people will just ignore it because most of us are too condescending?
Because there are plenty of people who are stuck and who would benefit from it. If they are willing to listen.
Learning python is somewhat off topic for this channel. Because getting a job is a highly social task. Few of us are limited by our technical capabilities in terms of which jobs we can actually get once we get fluent in Python.
For learning python, there are many many resources in other channels on this server and also countless tutorials.
Edit: this is my experience at least. If anyone feels that technical tests are a major obstacle for landing a job, then instructions as how to pass those tests is still on-topic.
I already outlined my "strategy" quite some time ago, I'm not rehashing it here
Ok I will search for it and add it to my list of examples, if I find anything.
But do I need to wait until I graduate to get a job?
Well, how do I know if I'm ready for a job. Is it by working on a project and stuffs? And do you need to be good at math and stuffs? Can anyone just learn to code
Just about anyone can learn to code, yes.
Pretty much every entry level job requires a degree in something like computer science, and you'll need to pass some upper level math courses for that.
Working on personal projects is supremely important for honing your skills, even if the projects aren't useful.
Also while at uni, apply to internships, those are pretty important, too
How much more much more difficult is it going to be for me to get a job without an internship? I have decent projects.
People's job seeking experiences have some many random or hidden variables that it's hard to give useful answers to 'how much' type questions
Fair enough.
There's a few basic things: having a clear resume (review it, get feedback), getting your resume on a managers desk (networking), showing more than just 'I went to school' (such as projects that show basic engineering skills, testing, GitHub, etc... even oss contributions), being prepared to interview: many people are unprepared for the easy questions like 'tell me about yourself'
I never did an internship, but I won a grant to do research as an undergraduate.
I recommend people do internships because that's the most plentiful kind of opportunity to distinguish oneself. But it's definitely not the only way
oh alright
guys
I have been doing an internship at an organization for the last 20 days, and the duration is six months. However, I have received a better opportunity. How should I inform my colleagues?
and i am foookin scared to tell them!!
You've been there for 30 seconds and you're an intern, don't overthink it. They don't really have a relationship with you yet, nor will you have become critical for any important work in that time.
Congratulations on landing a better opportunity 🙂
i am little bit scared what if they scold me etc..
I hear that, it's rarely fun leaving a role. I very much doubt they will, and if they do you're better off without them anyway. Wouldn't be necessary or professional in the slightest for them to do that
can i return there laptop and ghosted them which is easier?
Just make sure you give your boss a heads up before the rest of the team! It's courteous to let them decide how they'll communicate you leaving out
I wouldn't ghost them, no. The conversation might not be comfortable, but just ghosting them would be unprofessional and is likely to leave a bad taste in their mouths.
well let see!! will ask to my colleague intern what to do!!
That's a terrible idea. It sets a very bad precedent for your future career.
And if you ever were to come up in conversation, they're gonna distinctly remember you as that guy who just disappeared one day, which will kill your reputation with whoever they are talking to.
??
yeah!! will tell them in 1-2 days!!
<@&831776746206265384>
!clban 1248904981793214464 some kind of scam
:incoming_envelope: :ok_hand: applied ban to @prime kiln permanently.
Recently i have been interested in Data Science. And just started learning python.
So like what's advice from the data scientist who knows about this field. what we do, how and why?
real world experience, problems
When you get an answer share it with me too
how is the market for fastapi?
What market? Which geography?
I mean job market.
Ok so in which geography? For what level of experience? What kinds of companies primarily?
particularly in the UK and Ireland.
in my case; I know some python, build few small cli application. and I just want to start in any company.
now looking forward to deploy something with fastapi.
Man, it really boggles my mind when a company is trying for months to hire for a position, I get dozens of recruiters reaching out to me because I'm one of the most qualified candidates for the position, and then the company doesn't even want to interview me 💀
I'll say this, if you're trying to fill a role and it's taking you at least 6 months and maybe up to a year to fill the position in this market, you're doing something wrong
I think you should keep trying. for someone who is enough qualified for the job; I hope there is someone who reasonate with you.
I mean, one recruiter already went to bat for me and told them they should reconsider, but they turned them down.
Give it a few more months maybe then they'll reach out lmao
I hope by then I'm employed and can tell them they reason they missed out is because they screwed around 😈
I am recent grad in the UK, I have to land job before my visa expires.
I'm so sorry. I hope you're able to land one soon
Thanks, you’re good! I’d welcome any advice, project ideas, or suggestions about FastAPI. I’ll add them to my roadmap.
I hope so; You'll get employed by then.
Hopefully someone can chime in with some. Sadly my direct experience with FastAPI is lacking. Most of my work is in regulated medtech data/algorithm/ML. My generalist skills are not where I'd like them to be 😅
FastAPI is not going to land you a role by itself. I'd recommend working out what kind of role you want to target, and noting the most frequently requested skills from job postings for that role. That will give you a rough idea of the highest leverage skills for you to develop.
Thanks, I am leaning towards backend roles with python. Fastapi with modern development using llm and Rag is frequently mentioned.
Also, do you know which direction has stronger demand right now that tends to be more approachable for someone like me.
I’d say you’re doing great by sticking with ML and specializing in it, rather than being a generalist in many things.
I don't know about relative demand between different junior roles. What I do know is that unfortunately entry level roles in the UK are very limited at the moment. It's a rough time to be trying to break in. I'd recommend thinking about what networks you have that you could leverage. Alumni networks, peers, professors, clubs and so on... Networking becomes more valuable when the market is depressed.
The target I'm aiming for is deep technical proficiency in a core area and light general expertise across others for collaboration/reasoning sake. Although I am targeting more senior roles that are cross-disciplinary. I know there's a lot of first time job seekers here and I am not sure if that same advice is applicable.
T shape, pretty standard 🙂
Had a good day today- been at an SRE conference at the Bank of Scotland 🙂 Perk of the job!
Cool. I'm doing pretty well in getting far along in the interview process, even if they haven't materialized anything yet. I want to be careful giving advice that might not be transferrable/applicable.
It's very transferable guidance. T shaped engineers are a very natural shape, and it's also a pretty broadly recommended one. As you were saying, expertise in a domain or two, with sufficient cross-functional depth is a powerful combo.
Though granted, when it comes to juniors that shape can differ. On second read, I see you were delineating in terms of level rather than role. It's been a long day, I missed that the first time 😁
Haha that's alright, I get it. I've been spacey these days. Chasing opportunities while trying to pursue a masters and take care of myself has been exhausting and always feels like I'm failing at one or the other
You’re absolutely right. that’s why I’m working on developing a serious project that meets the criteria of someone with at least 2+ years of industry experience. You’re also right about networking. I’m reaching out to colleagues and friends, but they expect me to be skilled at something and to have already shipped a tangible project.
Yeah, it's health that slips for me... I used to be in shape, what happened?! Looking forward to getting back there come the end of my Master's.
Yeah, I think I've put on around 20 pounds since the start of my unemployment at least
It's easy to fall into the trap of "I'll worry about all of that when I'm employed" when you're in a long bout of unemployment.
yk, I've heard that for the future (or the now), one must achieve the π shape (or broader), i.e., become a polymath: https://www.youtube.com/watch?v=ZzhLI6gXgY0
General Translation, translate your entire app with 1 component: http://www.bigboxswe.dev/GT
Polymaths are the next generation of programmers.
Music
@massobeats : ruby https://www.youtube.com/watch?v=fpuogvBgPYg
umm, ignore the thumbnail?
Honestly feels like a lot of words just to present the T shaped concept with a different hat on 😁
I generally think of the T shape as encompassing depth in closely related areas rather than necessitating depth in exactly one (e.g. SRE, DevOps and infrastructure engineering are all fairly closely related domains). It's rare for it to be optimal to develop deep expertise in orthogonal domains, so I'd still recommend something like a T (or arguably closer to a V) rather than the π or comb shapes the video mentions.
Additionally, I've found that the majority of the time if an engineer appears to have a π/comb shape, it's a matter of me missing the connecting factors rather than their depth being in truly distinct domains.
On a different note, I've pinged you a LinkedIn connection request, @fast fossil . Wish I'd started coding as early as you did, haha
Grad school is absolutely rough especially when you get Imposter Syndrome. Having a good PI helps ease the pain but, if you have a bad PI....oooff.
What really is a PI
Basically your mentor/boss in your research lab, it stands for Principal Investigator
Does anyone have any advice about cryptography
What are you actually looking for? Advice in what regard? Cryptography is a niche within a niche. It doesn't pay that well, and there are very few roles working on it (many more implementing it, though that's very different)
I am interested in looking at being a cryptographer
I thought it did pay well
It's worth double checking me saying that, it's been a while since I looked into it so I could be getting my wires crossed and misremembering.
IIRC it pays well compared to most fields, but is pretty middling as far as tech roles go (albeit it's more maths than tech, I still lump them in the same bucket)
Yes i am interested in it because of the maths side of it
I also live in Australia which would change the pay i am guessing you don’t not many people do
As far as some countries go
100% it would, US and UK are the markets I'm more familiar with.
Yes i would be interested to learn what the pay difference would be
I would probably prefer to work in a research or government role
Post quantum
Easy to get a sense of that with a Google 🙂 Also look up how many roles there actually are for cryptographers. Worth being aware of that too, as your comp is 0 if you can't land a role
Ye lol
Ok thanks
What should I pay attention to when programming?
probably the code
Hi am currently pursuiing ece what are the options for me to do masters in it
hello guys im kinda new to coding/programming community... can someone help me on how i can start my journey... Id be grateful for your advice.
And once you're confortable with python you can start with novice project -> https://roadmap.sh/projects?g=python
If linkedIN is cringe than where to apply for Internships and Jobs ?
Maybe i should just use that jobs section and not the Home section
if you can, try to look for internships through your uni
Yes, using the section literally made for finding jobs might be a good idea...
me also bro today i study variable can someone help me on how we can start my journey.
What do you want out of Python?
That'd determine what 'journey' you should go on
i am learning for some new ideas and exploring
Hm... but that could mean anything
i think is should combine my carrer in coding and finance can you suggest me some about that
i am now 18 year old
hmm...
Err... okay first maybe try understanding Python fundamentals
Then you can work on finance-related (mostly and essentially) modelling on Python
can i add you as friend
I mean you can ask your questions right here on this server
I don't have too many exact insights to give about your specific sub-field, sorry for that
ok bro thanks
i know i new about this ishould try to understand people first
by the way there is any other discord community i should join
Hey guys, how are you. I've got a question, is there a certificate on ML that really matters?
Or not really? If it's a paid one but somewhat accesible then it's an option for me as long as it really worth it.
If you're wanting to get into ML, it's going to be a matter of needing a Master's+ rather than a cert
Understood. If that's the case, I'm already going to get a Masters, so then the new question is where can I learn ML in the meantime.
So I thought on reading this, but not sure if this is the best option.
I've done the Probability course already.
I'll let someone closer to the field than myself point you toward their preferred resources 🙂
@jade rampart you can check the pins in #data-science-and-ml (also a place to ask any relevant questions about learning the topic and such)
Okay, will do so. Thanks.
I completed variables... Just 1 hr ago
Where should i look for jobs in Data Analysis? I already use linkedin and some others but cant seem to find a Jr job.
This isn't the place for memes
the ending was fire
The job market right now is very competitive. What qualifications do you have for a junior analyst job?
i already worked for a medical institution as a data analyst
how can i learn python ?
!learn
Here are the top free resources we recommend for people who are new to programming:
- Automate the Boring Stuff — an online book (also available to purchase as a physical book)
- Harvard’s CS50P course — video lectures (slides and notes provided) with exercises
- Python Programming MOOC 2026 course — text-based lessons with exercises
- Corey Schafer's YouTube playlist
For a full, curated list of educational resources we recommend, please see our resources page!
How does one find gainful employment these days? That does not require LLM usage. My company just announced a "strategic partnership" with Anthropic and I no longer wish to work for them.
Recruiters, network, cold applications, whatever. Is your decision to exit the org related to the partnership?
Most definitely.
Why is that? If you're able to expand
It puts me closer than I want to LLMs and I fear that the mandate to use LLMs is closer than ever and I will not use them
What's your reasoning here? They're a tool like any other, I don't think a blanket refusal is a particularly rational approach
My ethics prohibit me from using something made from stolen data. My morals will not let me do something that contributes to great social and economic harm.
Its the same way as always, but you should know that youre making it much harder for yourself nowadays
I understand that I am limiting my options. But I will not bend on my principles.
I'd recommend fleshing out your LinkedIn profile a little more, it can be a great way to get inbound attention from recruiters without much effort on your part. It's largely set and forget.
Yeah...
Does not require or does not use?
Because plenty of companies don’t mandate AI. It tends to be the less sexy ones. The ones that don’t chase trends. That or really stuck in their ways ones that are a pain to work for.
Is the problem the partnership or general LLM usage
You'll probably struggle to find a job at a company that hasn't incorporated LLMs in their tools nowadays
Does not require. The place I am at is gung ho. Not mandating, yet. I think that the next evolution following the partnership will be the mandate (which I think we all felt was coming.)
The partnership. I am also not happy with how gung ho people who should know better seem to be about LLMs.
Ha, I'm with you there. I keep seeing people try and stuff them into places where a deterministic approach would be a much better fit. They're powerful tools, but should be reached for when they're a sensible fit rather than as the default first port of call.
hey guys! upcoming sophmore currently working on prsonal projects. whats the best possible skills i can have on my resume that tech companies are looking for, i am currently trying to expand my skills
Skill and ability to code and comprehend problems.
A specific skill ,,, well every company cares about different things. It is easier for me to tell you things to avoid than to tell you things to chase. Just follow what you enjoy and can dig deep into. Don't worry too much about the specific skill itself.
<@&831776746206265384>
!clban 845983263477465089 This is not a job board
:incoming_envelope: :ok_hand: applied ban to @peak fern permanently.
I'm thinking to take my btech degree in CSE quantum computing but just I'm too scared what if I end up doing nothing means what if I don't get any job and I will also be looking for internships during my btech what if I don't find any
And I'm also not very acknowledged about these things please can u help me out and if u can add some points I should care about
Do you think Quantum Computing won't be important in the coming years?
But everyone's working on it and if it goes well I think great opportunities there can evolve
What are you worried about?
Shall I go for it or not what if after 4 years till I graduate it's in low demand and I had to compromise with placements later blaming to take this course
What else are you going to do?
You are expressing doubt and worry, and that's understandable. But, ask a question that we can give an opinion or advice on: your choice isn't 'do quantum or nothing'
You're probably really wondering if we think quantum computing, or a degree in it, will be relevant in a few years
Is there anyone looking for a dev?
Right, because if there's one place it makes sense for a developer to advertise, it's to a group literally full of other developers... I swear, the lack of critical thinking from some people...
hi i am very confused like what to do actually
I have around three years, and I do not want to waste them learning things that will not provide long-term value. My goal is to become an AI/ML engineer, but at the same time, I also want to build my own applications and websites independently.
The problem is that I am confused about the right approach because the competition is extremely high, and there are too many different roadmaps online.
According to you, what would be the best path?
Should I first:
- learn a programming language,
- then practice DSA in that language,
- then learn full-stack development,
- and finally move into AI/ML?
Or should I focus directly on AI/ML from the beginning?
I would really appreciate it if you could suggest a proper roadmap, including:
- which programming language I should start with,
- which language to use for DSA,
- whether full-stack development is necessary,
- and the complete sequence of skills I should learn over the next few years.
I want to focus on skills that will still be valuable in the future and help me build real-world projects, not just follow trends.
The first step is learning a programming language. Python is the right language to start with. Get a degree, and worry about DSA during the degree, not before it.
How difficult are the technical tests during the interview process these days? Are they a major obstacle for any of you?
They are the primary obstacle during an interview process
How do you keep those skills sharp? Is Leetcode good?
Daily leetcode at least and more near your high intensity application days
Its much easier to keep some leetcode ability up than to forget and relearn everything every time you jump jobs
Serious “culture fit” problems tend to filter you out before the technical. So making it that far means your personality is at least close to reasonable and the non technical will likely be easy to pass.
We give interview questions and don’t allow them to vibe code the answer (but limited ai use is allowed) and I’ve observed a massive amount of atrophy. They are saying things that suggest they know or knew code. But their code ability is … missing.
So in that sense, technical interviews are harder for some people
Solve novel problems. Leetcode only gets you so far. But raw experience tends to be enough for me. And making sure I do non pure work code. Because sometimes work code is so specific that it doesn’t help me anywhere else.
Leetcode teaches you how to be good at leetcode. Some interviews are just leetcode questions and so that’s fine. But the trend is shying away from leetcode style questions.
-# but plenty of companies still use it
- most interviews are just leetcode questions
Unfortunately in the U.S. as I’m sure you’re aware of AI is pushed hard at C-suite as a cost saving measure. I found that my LLM use is too conservative when it comes up in interviews so I’ve begrudgingly tried to get more comfortable with it for certain tasks.
That said, since I’m in regulatory medtech it hasn’t been too hard to convince companies to avoid LLMs for anything mission critical, but I’m still seeing a slow ramp there as well.
That being said, a combination of the job description plus a conversation with someone at the company is usually enough to get a good read towards their feelings regarding LLMs. Will just be something you have to try to ascertain if it’s a sticking point. For instance, I always try to get a vibe check on how a company feels about complying with medical regulations and to put it directly what they do to make sure they don’t become the next Theranos.
Btw as a FYI from a discussion in pycon: https://pyfound.blogspot.com/2025/12/anthropic-invests-in-python.html
It’s actually been quite sone time since I had a leetcode interview, so I suspect it’s industry/scope/level based
I have a more nuanced answer than others may have on this.
Technical tests have definitely gotten way more difficult, to the point where I question the relevancy of the level of difficulty with respect to the job.
That being said, it's also something that can be prepared for and is at least predictable. Personally I hate the presentations on previous work and work philosophy and all that other nonsense stuff. I stress about that way more than technical depth of the position.
This may just be the type of positions I go for or my personal disposition, though.
I would be ecstatic to never have to give a 30 minute to one hour presentation on my past work/career ever again
I'd take a tedchnical assessment any day
That was a $1.5M sponsorship with no strings attached. That is much needed money. It is very different than the company I work for shipping a product that uses it.
<@&831776746206265384>
I want to learn python 😭
!learn
Here are the top free resources we recommend for people who are new to programming:
- Automate the Boring Stuff — an online book (also available to purchase as a physical book)
- Harvard’s CS50P course — video lectures (slides and notes provided) with exercises
- Python Programming MOOC 2026 course — text-based lessons with exercises
- Corey Schafer's YouTube playlist
For a full, curated list of educational resources we recommend, please see our resources page!
hi
Quick question, for a recent graduate or someone who is just entering the tech industry, is it recommended for that person to only choose jobs desired by the person?
For e.g, say I'm only interested in full stack development, does that mean if I have a job offer on another kind of development I should ignore it?
I was recently contacted for an interview for SAP development (Im still reading about what it is and what we should do) but what the person told me, it's also some kind of development. I already schedule by interview and I was wondering if it was the right choice to accept or I should stick only to thing I want like full stack development?
It depends on what you are willing or want to do? If you have time and you really need a job then take it, if you don't think you'll enjoy it then don't take it.
It also depends on how skilled you are, if you think you could nail a fullstack job then sure, aim for it, however, if you are just looking for experience then any job would be good.
yeah I see, at the moment, I'm mainly looking for experience
we are in same game i am covering with code with harry playlist you?
!warn 1363529222663962706 Please don't advertise your linkedin posts here
:incoming_envelope: :ok_hand: applied warning to @outer vale.
Job market is buring from last 10 years.
hi
linkedin is my favourite spyware :3
mossad agent reading how bob's divorce made him into the ultimate real estate agent:
well im seeing cs50 introduction to python
Hi Mates,
Hope all having a great day
I really need help
Idk i think i'm stuck,
I made projects to show case my skills and when i applied to internships i get rejected idk why
Also how can i learn more?
i'm feeling like i got stuck
I'm sorry to hear that. The job market is pretty bad right now, and internships reflect that as well. Did you interview for any internships, or did you not even make it that far?
In some. internships
they interviewed me and ghosted me
And others didn't even reply
What is the state of job market now
Its terrible and deteriorating
Is it deteriorating? I believe software jobs are up about 10% year over year?
Generally speaking it’s at a low and fluctuating. But still at a low.
-# some sub markets are healthy based on location and industry
Postings are up, are offers up?
Its a struggle all around
They can't be real jobs, the postings are just to collect data
What kind of postings they are is important too, if its senior jobs only who cares
print type shi
Where are you seeing that? I'd be curious to look into the source
is py useful with outdoors work? I'm chasing it up as a hobby / tool to automate stuff at home, but is there any other good applications for it?
employment programming very much seems to be a "can't get the experience if you can't get a job that needs experience" kinda deal
The job postings data is from Indeed
Gotcha, thanks.
I agree that postings don't match offers, but this is still promising. Should be noted though that a slight increase over last year still doesn't mean a strong market overall. Last year was awful, being slightly less awful doesn't make it not awful
Though notably, this was all in the days prior to Iran kicking off
That being said, I have noticed an uptick in opportunities
ooof, yeah
I really REALLY hate that everytime some catastophic political decision is made it just torpedos my job hunting for the next couple weeks as hiring chills and priorities are "re-evaluated"
I try not to get too political in not political spaces but it's hard to not when we see direct causal relationships like this
Existence is political. Agreed there's a time and a place to drill into it, but it's pretty inextricably linked with the entirety of our lives.
100%. And I've never seen it so directly relevant to the job market, but here we are.
I'm concerned that if/when the bottom falls out of the current AI train, it could have a chilling effect across tech.
That's a valid concern, but I'm not sure it will be the case. I'm not super confident in this data point or anything, but IBM reversed their AI initiatives and started hiring entry level tech employees. Would be nice if that became uniform.
We shall see, and hopefully so.
I do still think the outlook for tech is extremely strong in the mid to long term, whatever happens in the short term.
The AI stuff is a huge problem
are we going to go back to the days of 3month long bootcamp got you 6 figure job?
My organization is fully AI driven and it's a complete disaster, we're probably gonna shut down before we decide to stop vibing
I will never be confident about future human behavior because I think assuming people are informed and behave rationally is a losing bet 
Hmm? That's about the reverse of what I was saying
man can dream, no?
Maybe the reverse, there will be a hiring surge to fix all the slop every org has adopted
Every org will hire a Chief Refactoring Officer
Heh, for sure. And actually, on that note... I know I've been thinking about my career too much recently, I had Google SREs popping up in my dreams last night 😆 I wish I were joking
Software Restoration Engineers
Desloppification Specialists
guyz ,do u know some good resources for Adaboost ??
what percentage of w11 is AI written rn? 30%?
What do you mean by w11?
windows 11 (probably a bad abbreviation lol)
Well
You see the problem with Windows 11, isn't that its vibe coded, but that it uses the same base code from windows xp
I feel like "AI written" is kind of a misnomer. For example, I just did a code refactor. I knew everything that needed to change, the tests I needed to introduce, and the end result.
I wrote maybe 100 lines of code, then told an agent to do the other ~1500 lines. I reviewed everything it wrote, made sure it passed my tests, and nit-picked where I wanted to. Does that mean it was 93% AI written, or did it just save me 30 minutes of work?
A lot of people (executives, business owners) see that percentage and just assume the developer only contributed 7%, then fire them lol
I get the impression a lot of AI companies make no distinction between the above case and unrestrained vibe coding on purpose, as a marketing strategy.
hey i think you all need to go in ai and data science
I'll stick to my happy niche in DevOps/SRE 😁 AI might be on fire at the moment, but it doesn't remove the need for engineers with other technical skillsets
good but not enogh
You don't think DevOps/SRE is also exposed to AI automation?
May I ask what that opinion is grounded on? Do you have industry experience?
i was like to do same question
It's not a loaded question. I'm curious about your opinion.
yes full experice , now i am building a llm that is powerful with 2 trillion parameters but i can run on simple laptop
yes , my opinion is to go on ai because ai companies are hiring ai developers
your curiosity is right on the way
I was asking @solid parcel
ohh i did not understand sorry for that
escape room stuff can you tell me about you
@vast shoal can you say to escapestuff to reply my question
Chill out, they are clearly typing a long response.
Yes and no. There's certainly a lot of places I think it can contribute. In practice, we're a long, long way from that meaning that engineers will be replaced. I was having a conversation with an ex-Google SRE and current dev advocate last week. Google have been experimenting with using AI to help with incident response. I.e. investigation, implementing generic mitigations, writing up a post mortem.
They were advising not to use it in prod because the agent is prone to misdiagnosing, and making things up. On top of that, frankly a lot of orgs don't have the structure in place that I think they need to leverage LLMs most effectively.
Humans are prone to failure. So are LLMs. We've spent decades learning all the safeguards we have to put in place to allow engineers to move fast, and with LLMs a lot of that good practice has been thrown out the window when it's more important than ever. Change increases risk of failure, LLMs increase speed of change. Ergo, LLMS increase risk of failure. I think we'll see a bifurcation in LLM impact between organisations with highly mature engineering practices who have the processes and tooling in place to leverage them safely, and those that are working to integrate them without that layer.
tl;Dr We need to treat LLMs like employees
no the jenson heuange says ai not replace person from jobe ai creater replace thats why i am working on a project name nexora2 thats an llm
Plus frankly, even if LLMs were capable of replacing every DevOps engineer and SRE on the planet, the way I see companies move I think it would take them a decade+ to manage that.
Sounds like it's basically the same as for development.
yes it sounds
I would agree, yes. See AWS rediscovering code reviews after LLM related incidents, lol
we are not doing and fun here i am sirous
what are you typing demontati
I leverage AI a lot to speed up various processes, but I always have human review checkpoints for important decisions, like merging, deploying or directly updating prod environments. Basically, if it's something with irreversible consequences, I ensure it's a decision I the human am willing to take.
I'm being serious, I'm laughing at the absurdity of Amazon having caused failures by treating LLMs like magic. They ignored engineering practices we've had in place for decades, and it caused them massive issues because they were giving LLMs too much trust
you are right
yes thats a funny thing
It's quite surprising large and mature organizations like that would end up in a situation like that. I never worked for a large enterprise like that so maybe I'm just unfamiliar with the processes at work there.
Hahaha, count yourself lucky 🙃 A ton of orgs (big, big ones at that) are still sshing manually into boxes to fix issues or make updates. One by one...
just saying, but you could physically not store 2 trillion parameters at even a 1-bit quantization level on a typical (simple) laptop (without additional compression)
EDIT: nvm, I'm off by a factor of 10: #python-discussion message
That’s a really sharp observation. Large enterprises often look polished from the outside, but inside they can be surprisingly messy. The sheer scale of operations means processes get layered over time—sometimes decades of legacy systems, policies, and cultural habits. Even with mature governance structures, things like bureaucracy, siloed departments, or outdated workflows can cause breakdowns that seem shocking given their size.
In fact, paradoxically, the very maturity of these organizations can make them more vulnerable to inefficiency. For example:
🏢 Bureaucracy creep: Decision-making requires multiple approvals, slowing down responses.
🧩 Siloed teams: Departments may guard their own data or priorities, leading to misalignment.
🕰️ Legacy systems: Old tech or processes are hard to replace, but still critical to daily operations.
👥 Scale of workforce: Thousands of employees mean communication gaps are inevitable.
Smaller companies, by contrast, often move faster because they don’t carry that weight of history. But large enterprises trade agility for stability, which sometimes backfires when unexpected challenges arise.
!rule 10
And you know this already, you've been told off before
You should remove that message
One of the key reasons I don't see them replacing human engineers is their inability to properly retain information. There's so much background context that goes into a lot of engineering decisions, that LLMs can't currently retain. Broad awareness rather than simply the specifics of the particular issue at hand... I've seen that cause real issues for LLMs, particularly when working on a non-trivial codebase, or designing a process that works not just generically, but in the context of a specific organisation
i was just to say what ai says about your responce
!rule 10
It was already explained to you why that's not permitted. If anyone wants an AI analysis on the discussion, we can do that ourselves. You're just creating spam.
I'll ping a moderator the next time this happens.
Yeah, I agree, and that's why they sit firmly in the "tool" category rather than a real stakeholder.
Exactly that, yup. Where I think my current org could by making wayyyy better use of them is for ticket deflection/triage/troubleshooting
A lot of the issues that hit our service guys are well understood, duplicates, or not even a fault at all. So being able to reduce the amount of human cognition required by having an extra layer of validation via an LLM would be very useful, and I do think that's something LLMs could do effectively. It doesn't have to be perfect to bring value, and this would be pretty low risk
hey all you can help me now
They are super useful for information acquisition. I find that most of the time the information I get is correct and useful (particularly since the Opus 4.x generation that was released at the end of last year). Almost any use case that involves finding, retrieving and integrating information from various sources is viable if you provide the right interfaces for it. It's just not reliable to have them act unilaterally on that information when it really matters.
yeah ,but i changed this impossible to possible
"reduce the amount of human cognition" doesn't sit well...
what are you talking about boo boo
"Reduce human cognition" is a bit misleading in this case, it's more like freeing up human cognition from mundane tasks so it can be reallocated to where it's more useful
One of my absolute favourite things to do is use them to trace through functionality in a codebase I'm unfamiliar with. They've already helped me find a couple of bugs in an OSS tool I'm building on top of.
I also agree with your reliability point. My preference would be to extend to them a set of deterministic tools they can use, that we're confident they can leverage without causing issues. (Which again, is not dissimilar from how we treat employees and least privilege)
ok but i not know about this topic
yeah that's more accurate lol
Exactly this, couldn't have put it better myself 🙂
this smells like slop
what smell
it looks like slop
what do you meen by slop
😭
:incoming_envelope: :ok_hand: applied timeout to @jolly sinew until <t:1779533358:f> (10 minutes) (reason: duplicates spam - sent 4 duplicate messages).
The <@&831776746206265384> have been alerted for review.
@neon pier please stop spamming the channel
what spam now i not spammed now
Only two hours late to the party 😆
if i wanna do ai things such as ml does C++ worth it?
there's so much theory you have to learn for ML that the only programming language that's worth your time is Python.
you almost certainly won't use C++ for ML at all, so all the time you spend learning it is time you could have spent learning something you will use.
if you feel like learning a high-performance language anyway just for the joy of learning (which is a valid reason), I'd suggest you learn rust. But that almost certainly won't help you with ML, either.
got it tysm!
hi
hi
Good day guys
Any advise for a beginner hoping to be a professional cloud engineer
And can I get a partner I can sharpen my skills with?
I want speak with the founder
the founder of python, or of this server?
Yeah
Guido founded Python but he has nothing to do with this server. If you want to talk to one of the founders of this server, let me know what you want to ask them and I'll take a message.
Check DM s
but your bio say’s you are founder
No in another sever
Is a Python v2, They've let me do it; it will be like it's oriented on the second server of this server.
alright
..
To be clear, we have no affiliation with what David is doing, and neither we nor they have an official relationship to the PSF.
this channel is for talking about careers, so make sure every message in this channel is about that. Advertising other servers is not allowed.
Are there still normal python jobs available or is everything ai and vibe coding.
where did you get the idea that people with no actual programming knowledge are getting paid to vibe code?
maybe they mean python jobs that would have otherwise been "manual", now being automated away through agents or surrendered to vibes
gotta say i was "chastised" or officially unofficially told off for not using enough AI, whatever that means
its possible they were hinting i was slow with my tickets so you never know 🤷♀️
But people who do have that knowledge are still getting fired ??
It's massively cheesing me off that companies keep using level of AI usage as a metric for success, it's completely inane. It's a means to an end, not an end in itself. It's tantamount to insisting engineers use as many servers as possible.
Yeah. Do you think they're getting replaced by vibecoders?
Is anyone new here?
A'm new and I feel it well be fun to learn together!
What about agents, memory, readme, mds, etc. to help w the context?
Hey guys what do I do if the founder has unrealistic goals that cannot physically be achieved and does not take any feedback about the viability of the business plan which is doomed to fail. Do I just ride it out and look for a new job?
This doesn't solve long term learning; that information needs resurfaced every single time and quickly fills up context. Models have large context windows now, but performance degrades rapidly long before those are even full.
If they're not listening and are dead set on something that you think will make the business crash and burn, then yes I'd be looking to jump ship. I wouldn't quit without a role lined up, but I'd certainly be applying for roles ASAP.
Are you able to share what you feel he's being unrealistic about?
I don't want to disclose the details but the essence of it is he's fully taken the AI pill more than you can imagine
It's actually very instructive to write things out like this, I get stuck in the minutiae of the day to day but when I spell it out this plainly the situation is obvious. Life lesson
Writing and rubber ducking is great, as is taking a walk and sleeping on decisions. I honestly spend much more time thinking and writing in my role than doing, lol.
It's often more constructive in the long term
Writing code was never the hard part. The hard part has always been figuring out what code needs to be written.
Honestly in my role it's often not even a matter of what code needs written, but what direction I need to nudge organisational thinking in. Trying to steer us away from some anti-patterns in process and thinking, while keeping the people whose work I'm advocating we tweak onboard. Wooo, politics...
That too. It's steering the ship and solving the problems. Explaining why the "great idea" isn't actually a viable thing.
Heh, yup. My org is trying to do SRE, but they're a big old fashioned bank. The short version is that the way we're going about things, SRE will end up being a policing function and a blocker rather than an enabler for dev teams.
I want us to shift from a consultative model, toward a product approach where we're building out tooling to make it easier for our dev and infra teams to do SRE well, rather than them having to rely on us. 80% of the stuff we should be able to handle via golden paths, the other 20% of edge usecases we can continue to consult.
The joy of management that has little to no technical understanding. I feel your pain.
Yeah 100% agree. I mean imagine if in dev interviews you made them do a typing test and were like well this guy had 100 WPM and this other guy only got 50 so he would only be half as productive
Please don't give them any more stupid ideas. Live coding interviews are evil.
They show literally nothing about a candidate's capabilities.
Happily, my boss and skip level both trust me. Unhappily, I've only just joined this team (from another in the bank) so my political capital is low, and it sounds like my skip level may have his eye on retirement. If that's the case, he likely won't want to take any big swings so that he's leaving more of a blank canvas for whoever comes next. So may be in a holding pattern for a bit.
It sounds like you're trying to prevent a big whiff.
Yeah, and my sense is that it's wider than my lab and platform. We had an org wide SRE conference last week, and my impression is that we're not doing a lot of the basics very well across the board
Not too surprising in a bank. I've hit that in engineering companies and that's really frustrating.
Yeah, my org is very good at making it very hard to get things done productively. It's ironically so restrictive that it often adds risk rather than removing it. Everyone is aware of the problem, but the residual culture means that teams keep inventing new blockers at the same or greater pace as simplification initiatives remove existing ones.
Slowlyyyy changing, but a lot of great engineers just get fed up and leave, so the culture is changing at a glacial rate
It is very hard to change culture.
That is sadly exactly what I'm running into, yeah....
"That's the way we've always done it!"
The painful bit is that in this case, they don't seem to see that they're making the same mistakes that have bitten them before
Unfortunately most companies and people don't actually learn from mistakes most of the time.
I will switch my major from data science to computer science because I can learn different areas. Is my case possible because I studied economics for a decade? I hope to study well in my school! Please give me your advice!
Why wouldn't it be possible?
This is my first time to study computer science!
OK. Everyone starts sometime.
How to get My first job in the 2026 era ?
I have never study computer science any recommendation?
apply on linked in
recommendations for what?
Hey guys im looking for experienced python scripters which are also UI designers
Applied But not get any Recommendation or Reply 😕
Do you want a Data analyst? Worker
no im currently developing an app
Are your major computer science? You are smart!
i'm still in HS, and I am thinking of pursuing a CS major (or any tech-related field that involves computers, programming and electronics and whatnot) when I go off to college. i have thought of the following questions:
-
how good/bad is the current situation on picking CS as my major? will the market improve/decline/change significantly in the next five years?
-
does the college I study in matter for finding a job? what extracurriculars/side projects would help me increase my chances of getting accepted into a CS major in uni?
-
if CS is not viable in the long-term, what alternatives would there be?
The first question is whether we should consider the updated curriculum, such as AI in cs programs. Also, the good cs programs should teach us different areas. If we learn from the updated curriculum, we should find a job.
The second question is about our programming skills. If we have a strong background, we should be accepted into a university.
The third question is whether the CS program is changing due to AI. You can consider double e because double e is not outdated.
so one of the universities in my country has a subject like "intro. to AI", would that be a good sign that the degree is worth taking?
second, what defines a 'strong background' ? could that include some projects posted on GitHub, and a self-hosted website (which I'm planning to do)? will they also consider how many languages I have learned?
-# i forgot to mention I still am quite new to programming in general
The first question is yes because AI is a new area for cs. The second question is yes because your projects in a programming language can state your abilities. The third question is that we are beginners, so we need practice and training.
I've always felt "strong background" was very strange to care about in a new-grad. You don't have a background other than having grown up and gone to school, what does a "strong childhood" look like that isn't a violation of some kind of anti-discrimination rule?
"strong, visible effort during school" is all that's ever mattered to me as an interviewer.
Obviously your mileage may vary, perhaps industry has made up some kind of definition of that, but if so I've never run into it.
I hope it's not just code for "went to a super expensive school."
Are you an interviewer? What are the things that improves the chances of people getting hired? i'd like to know from an actual interviewer!
I have been many times, I retired from my 28th or whatever year as SWE last year.
For me, the top thing has just been asking the candidate to ponder some question aloud, "showing their work"
You can tell right away IMO whether they have what you're looking for with that, even from a "simple" question like: "OK, you're on your choice of computer, your choice of web browser, you've typed https://example.com/ into the URL bar. You're hitting Enter, and the computer has a working internet connection of your choice. What's everything you know about that takes place next?"
I've dis-recommended a zillion people after that, A+ reviewed some, and proposed up-leveling a few of the best.
One of the latter is now a pretty senior engineering manager in Google's adtech team.
I hired him in 2008 when I was "Principal Architect" at a defunct social media startup in NYC.
I feel like the skewed supply-demand curve of the current job market has made recruiters and job ads start adding bizarre requirements and expectations of the applicants.
Do you have any examples of what a bad response to that question is like?
Anything that starts with the remote webserver is an instant fail from me
The best folks start by talking about the keyboard switches
My favorite insane one-liner question is: "You're a UNIX process on any system since 1970. What's file descriptor 2?"
and then you watch their face and say nothing
Ohh thats so creative!
If they ask any questions, they fail
That's stderr?
(only applies to systems / SRE / etc roles, not CSS people obviously)
Yeah
You pass! 🙂
Anybody who has ever done find . -name foo 2>/dev/null and doesn't know that answer is a cargo-cultist.
Man i dont even know enough about windows xp 😔
What do you mean ponder questions aloud showing their work? Talk about what they did?
From the day Windows 95 came out, I installed it, fired up regedit, and deleted everything I couldn't understand. Reinstalled as needed until I had a working system.
No, just explain their thought process of working through the question
Harsh
My Windows 98:SE install was 40mb on disk.
Are all interviewers like you?
I know what every relevant thing in the Windows registry tree does now
Luckily for most, definitely not.
I'm super crazy
That seems very unlikely
how did you become an interviewer? What do most interviewers expect then?
Most are about "culture fit", and "STAR", and "architecture interview" etc
Once you are "staff engineer" or higher, you're always one of the interviewers
Oh
If you're a senior engineer, it's pretty common to be asked to interview applicants
to the point where it's a thing you have to push back on or they eat all your time
Only on a question like this though to be clear, where I've posed it in such a way that it contains all the information I would give them
re: "show your work aloud", when I was interviewing at VMware for the team that became Cloud Foundry, they gave me this whiteboard problem of "Knight's Tour", a chess problem I had never heard of, and I wasn't a chess player.
I told them that, and said I'd give it a shot, and just narrated my thought process at the whiteboard.
I feel a little ambivalent about a question that checks if you have some singular fact memorized or not
They offered me the next higher seniority level above what I was applying for, and I ended up writing the API server and Services API for Cloud Foundry solo
It's not a fact though, it's the way the whole thing functions
Meta uses that question in their SRE phone screen, hard fail if you don't get it
that stderr is file descriptor 2?
Yes
what was your thought process?
Can't exactly remember, but basically it was about constraining the number of squares I had to evaluate, and not having to revisit ones I'd checked
apparently I got it right, I looked it up later on Wikipedia, but they gave me a Staff Engineer role instead of Senior SWE II
Could you explain how that's the case?
This was like 2011 or something so it's not that fresh of a memory.. but they turned it from a half-day to a full-day interview and had the former Sun kernel team interview me too
OK, you could start with "have you ever redirected warnings to /dev/null?"
and THEN if they say yes, it's a hard fail if they don't know what "2" means
If they haven't ever run a redirect command, then you don't want them for your systems eng team anyway etc
It's a "cargo cult" check, not a trivia contest, I guess is what I'm saying
So you mean it would indicate they don't know the syntactic structure of a common terminal command to for example redirect stderr to stdin, etc?
There are a lot of similar questions you could ask, I just like the brevity of that one
No, I mean that if they've ever typed 2>/dev/null and don't know that fd 2 is STDERR, they just copy-pasted that idea from the web and didn't understand it
and I never would hire someone in charge of production UNIX systems or the code on them that was like that
understand the command before you hit Enter
I would let them make up the question even honestly
I don't really care if they "know that one", it's more of a way to check how they approach problems
It's a privacy concern so you can't ask, but "type history on your shell for me" would be just as good
Right, that's kind of what I meant as well
I don't care if they are a shell wizard, but I want to know if they understand what they are running
They either don't know common terminal operations or they've run them without understanding them
Yeah
Meta also asks some more "trivia" questions IMO like "how would you see what syscalls a program is making on Linux"
my answer was "strace, but I don't use Linux, I use FreeBSD, so I'd use dtrace or truss"
They really wanted me for their SRE team after the phone screen but I refused to work for an adtech company and politely declined the on-site
but they had some fun phone screen questions I copied for later gigs
they were Facebook at the time still
the New Relic interviews were irritating; I passed, ended up as "Polyglot Engineer", a role they made up for me because more than one team wanted me, but the Java one was dumb in particular, all nitty-gritty did-you-read-the-release-notes crap
@pastel aspen you are working in a company right?
Not anymore, I have my own
@pastel aspen just trusted my insticts and did this. Do they expect to solve the full board?
"Compiler Admiral" was my favorite title, Engine Yard gave me that when they paid me to work full-time on the Rubinius (Ruby VM in C++/Ruby) project.
It was whiteboard, so they didn't make me "execute" it, just write the algorithm out
You are a systems dev? Is there demadn for such jobs?
Yes, and yes. I have a ton of advice about that as well.
Everything "web" is pretty trivial to "agentify" now.. Systems, particularly "hard real-time" ones, are where you need humans the most, and always have.
that sounds simple. backktracking should work right? And I like programming in rust so is there a demand for programmers who know rust?
IMO learn Zig, learn Lean4, learn Julia if you want to work with math/physics people as well.
Rust is good, I prefer Zig because it's more of a straight razor, and comptime is better than Rust macros
@pastel aspen how can I get My first job in the 2026 era where the competition is very very tuf
Rust is pretty good enough for physics and maths is it not? Im building a physics engine currently
But the Rust ecosystem is huge, lots of jobs
Find a cool company that is "deep tech", reach out to people on LinkedIn/Reddit/etc that have previously been interns, ask if the culture doesn't suck, directly contact their CTO/CEO if the answer is "they don't suck"
In what ways is zig better? Which langs do your company's codebases use commonly
Really?
Oh yeah, Oxide Computer for example is a great place to work, all Rust, all open source, etc
should i take time to learn zig or is knowledge of rust enough?
I only use Zig and Lean4 currently.
It depends on your intent. Rust is great for getting jobs, Zig is the frontier for deterministic hardcore systems programming.
and you have experience building vms and stuff like that right? Why is zig preferred?
What do you mean "they don't suck"
Rust has "macros", Zig comptime is just arbitrary Zig. Zig is a drop-in compiler for C, and a badass one, so you can migrate a project or just merge C and Zig very easily. Zig has no warnings, all such things are either documented and fine, or hard errors that fail compilation. Genius. Infinitely fewer cases to test in your Continuous Integration setup.
Not abusive, not hardcore drinkers at the office, don't force you to attend improv classes (literally a thing I endured), etc
Just ask for horror stories, and if there aren't any, you're good
One place I worked had a rolling cart of hard booze that people drank all damn day, awful
Several others have had ping pong tables in the open that people played all the time
and it was not permitted to ask them to chill while you are on a video call with a customer etc
damn. maybe il learn zzig later
Code Climate had a super loud Pac-Man touramanent arcade machine in the open office etc haha
Seems kinda fine except for that detail
Zig is semantically also far less-complex than Rust, the Rust compiler is inherently gnardog
Exactly, it's why you need stories and not just bullet points
Former interns are great because if anybody is going to be abused, it's the interns
The "principal engineers" might have nothing bad to say, but maybe only because they were treated like gods
the top New Relic guy had a glass corner office filled with his guitars etc
Rust helps you to make what? Application or website
but the moment I needed sick leave, HR went apeshit on me
I would say applications and system stuff, not websites, though you certainly can
Rocicorp's Zero is what I'd personally do websites in instead though, were I someone who tolerated JavaScript
If I were anyone else, that would be fine, but I'm crazy and after having been on the NodeJS team at New Relic, I refuse to use it
Give her Tight slap 👋
Once you've read the V8 source code front to back, you come to have a different opinion
I just quit instead actually, I don't tolerate companies that tolerate evil HR
I've done a lot of quitting
Bro what
Not sure what you're implying by that, but it's probably not ok on this server
@pastel aspen Whats the msot common type of work to do in systems jobs?
Yeah, I would edit/delete that, not what we want here.
Honestly, debug other peoples' stuff haha
Accidentally entered before completing
(I meant LEGEND's thing)
What is the "stuff"? Like which type of things do you develop a lot?
The thing with "systems", and why I don't personally use the word, is that it's everything
Some people mean "trivial UNIX daemon", other people mean "thing that happens when you move the joystick in an F-35 II Lightning"
For example, you might be at a company that makes network switches, and now you're writing more or less firmware compared to the folks who are "just" making a db with Apache DataFusion
and you don't get "print statements" to debug a network switch
Oh alright
So it's a little too broad to say; it depends completely on the company
and even then, there may be multiple levels of "system"; for example some people mean Kubernetes, I would never call that systems programming but lots would
also almost everyone ignores the distinction between real-time and what I call "batch jobs", aka everything you might do on Linux or a web-app
Whatever you’re saying is going completely over my head. Your knowledge is seriously impressive, wow!! "Defiler" this is how Experience person look like
Learn ZephyrOS, QNX, or VxWorks, IMO.. The three most-popular real-time operating systems
Zephyr is a Linux, the other two are different, both amazing designs.
One of these runs every NASA craft, every nuclear power plant, etc
I literally don't have any powered-up Linux hosts in my life, never did, only did at work when they paid me to care about that
It isn't that I don't like Linux, it's that it can't do anything I want to do
Learn what a PTP Grandmaster is, for example, cheap to have your own at home
The moment you have that in your world, your perception of "done in time" changes forever
The PCI-Express bus cycle time is way too slow to handle PTP
This is why I say most things people call "systems" are actually "batch jobs"
What is it you want to do that you can't do on Linux?
Anything real-time
Linux has no facility for hard real-time
So, if you need to "keep up with" a sensor in the real world, gg nextmap
Same reason you can't have a garbage collector, you can never "pause."
The F-22 is written in Ada83, the F-35 II is (ugh) written in C++ instead, but they kept the F-22 avionics core because it would be insane to rewrite and re-prove that.
If you have a GC pause when you're trying to manage the state of a nuclear reactor, things don't go well
For me, it's vehicular sensor fusion applications right now, and I don't want to crash because my "ego-motion delete" 2000Hz loop paused
I can see why you'd need that for certain applications, but do you often build F-22 software in your spare time?
I'm writing a Zig library for PT-CAN to replace the software in my car's transmission
Yes, effectively
My point to this ramble is that "certain applications" are the ones that pay $700k+ a year right out of the gate
You put the software you write in your spare time in vehicles you drive yourself?
But if you want to grind out the local maxima of webapps until you're my age, go nuts.. but my advice for new folks is to leave that to the AIs and aim at the harder stuff
Absolutely
skin in the game
It's how you "git gud"
"I will die" is a great motivation for QA
That may be one possible outcome, I suppose
I have an "iron bird" set of full components of the car on my bench, exact modules, exact firmware
duplicate of all the electronics, same acetal ground plane sled, same thermal plate, etc
(iron bird is Lockheed Skunkworks' term for a non-flying duplicate of the vehicle etc)
Fly what you test, test what you fly.
Measure thrice, cut once. Overkill is also kill. Seconds count.
😵💫
I have never had to review a bugfix PR that involved code I had put in production.
I write tests that exhaust the parameter space of the inputs to the program, and if I can't, I delete features until I can
@pastel aspen how many years of experience do you have? In computer field
This is my 41st year as a programmer
Seems like a reasonable approach for vehicle software.
I retired with 28.5 years of professional SWE
Actually I guess I'm in my 42nd year of that now
I got a tutor for 6502 assembly on the Commodore 64 for Christmas when I was 6
next language was LOGO
(I tell people to learn from Turtle Academy these days no matter what their intended language is for work)
When the NES came out, and we all got them, I already knew how the games were written
If you want to Advise me or others what we have to learn in 2026 cause you have 42 years of experience in swe
I learned by typing in game code from Byte Magazine, fixing any typos I'd accidentally entered, and then trying to make some trivial change to it
I literally am, above, honestly, glad to elaborate
but my advice is Zig, Lean4, Julia. Learn a real-time OS. Set up a cheap Precision Time Protocol server and plug it into your computer.
Never let an LLM write code for you, you don't gain mastery in things you delegate
Remind yourself of this every morning. Read or write something, even something super short, before you let yourself look at your phone in the morning.
"Simplicate and add more lightness."
Never permit yourself to "pick an adjective" for what kind of Engineer you are.
unless it's "Physics Engineer" or something totally all-encompassing
Always try to predict, not react. Always look for a silver lining in anything that doesn't damage a future capability.
I have Taken the Screenshot Of your Advice thanks 👍
The masters are men, you are a man, if you think you will be inferior to the masters, you will be on that road very soon. (also Tsunetomo Yamamoto, forgive the 17th-century gendered language)
"If you can't explain it to a layperson, you don't understand it yet", "Admissibility before Optimization"
Always be reading a book that's "too hard" for you.
read Unfettered Mind, Hagakure, Dao of Jeet Kun Do, The Mind Is The Final Weapon
Almost all textbooks are awful, treat them like "tables of contents" that lead you to actual investigation
Don't be put off by awful terminology, it's everywhere, just keep at it until you figure out what they should have called it
Here's my current "reading list" (pool really, it's not a stack or queue)
@pastel aspen my current skills are Python(numpy pandas matplotlib), Power Bi, Machine learning(Sk-learn), MYSQL, Excel, Weka For Machine learning, Theoretical Knowledge of data warehousing and big data
Cool. Try reading some random Weka modules, see if you can fully understand what they do/why they exist
Made 10 Projects also
Oh, and go read what "newbies" are asking in the domain you are learning, and only chime in if you feel like you've figured out how to help
That's what I did here to brush up on Python, and they made me a helper about as fast as it's happened, apparently
I've shipped like.. one? total Python thing ever (it worked well)
and I don't use garbage collected languages anymore so I guess I won't be doing more, but I still think Python is a great thing to learn, and I'm glad to be here etc
Read the Python eval loop C code actually, it's very good
Elite C
(re: C, check out Deep C Secrets: Expert C Programming; old but full of amazing things)
Has a fish on the cover
Peter van der Linden, old-school Sun
@pastel aspen What kind of projects should I be doing?
in low level langs
Anything you are passionate about, especially if you can arrange for "skin in the game" to motivate you
re: that, read Antifragile by Nassim Taleb, but ignore his personality, it's.. not great. ..but the ideas are crucial.
"Resilient" is surrender; you only can stay the same or erode.. "Antifragile" is what happens when you train at the gym, you get stronger from stress, etc.
I cant think of any skin in the game type projects 🙃
Some ideas maybe?
What non-programming hobbies do you have that you might "enhance" with computers?
e.g. if you play tabletop roleplaying games, make a Foundry module, it's super easy, their API is great, etc
None.. Just robotics and computers.
OK, then I would go find some hobbies first and then program for them haha.. is there a "maker" space in your town? You could learn to write G-code for an embroidery machine, 3D printer, CNC mill, laser cutter, etc..
learn to tie knots and then try to write a Python program that can figure out whether a photo you took of one is an actual knot, or a "chain sinnet" that you could pull on both ends of and turn back into a straight cord
(topological data analysis, Betti numbers)
Go get an MRI scan and ask for the raw k-space binaries and write a program to process them
or PET scan and the list-mode binaries
G code? I cant even build desktop applications properly 🙃
You can never do things you've never done before until you do them.
Is it tough to write code for embedded systems?
"if you think you will be inferior to the masters, you will be on that road very soon"
They let you have that if you ask?
Liek i have a couple of chips like arduino, rpi, nano, esp and all those. Maybe i could try programming some firmware??
Usually the tech can't, but there's an e-mail you can contact etc, depends on the provider
Niels Heinrik Abel, one of the best mathematicians ever, when asked how he'd done so much in just six years, said: "It seems to me one should study the masters, not the pupils."
French mathematician Charles Hermite said: "Abel has left mathematicians enough to keep them busy for five hundred years."
Abel's work is what powers the elliptic curve crypto we use for all e-commerce now
@pastel aspen bye bye thank you so much 42 experience person 🙇
How did you get into programming firmware? how do you do that?
You just read the code for something, modify it, see if you can not break it, etc
You can't just "first principles" write some for a device you didn't also design, IMO
What if the code wasnt open source
Decompile it, fun learning experience
(the 386 microcode just got decompiled, super rad)
learn to use Ghidra etc
also, there's no need for non-open-source
Do yoou know how to use that? And is it tough ?
Get a Lattice iCE40 dev kit, super cheap, totally open-source-tooled
Barely, and yes, but for people who "do that" it's really great
"When you learn to read assembly, everything is open source"
Whats that for?
Very true, and that's why I thank my lucky stars the weird fellow EE student my dad got me for a tutor chose to teach me 6502 asm at age 6 instead of BASIC
Field-Programmable Gate Array; literally for anything you want
like an arduino on steroids?
there are also Field-Programmable Analog Arrays, e.g. I have a bunch of "Renesas AnalogPAK SLG47004"
lucky you. i self learnt everything from 10
Hmm.. Not exactly. More like an Arduino where the hardware can morph like a superhero
It's all SRAM and "programmable wires", basically; SRAM is how the cache memory on CPUs works
It's very different from Python etc but there's a badass FPGA next to every quantum computer you've seen a picture of
The thing in the photo that people think is the computer is the refrigerator, in the background is a "boring" stack of ultra-high-end "normal" computers including a (usually Xilinx brand) FPGA, because you have to do error correction and control the microwaves they all use to read the data etc
it's exactly the same device as lives in every 5G cell tower, down to the model
and those 5G towers coordinate with the PTP (Precision Time Protocol) standard I proposed putting on your desk
that's how call handoffs work as you drive around
GPS is approximately the same idea
Also, learn biology and photonics, don't stick to electronics
The first two are where the future is, I call the latter "chargetronics" because electrons actually have two more-interesting "degrees of freedom" than the charge/voltage we use right now
spin, and valley
spintronics is here already to an extent, valleytronics is coming next
magic angle graphene, tungsten diselenide, etc
inverse faraday effect can induce semiconductor modes in graphene with photons
Akhetonics in Germany has a working all-photonic general purpose computer
BrainChip in California can do 60 watts of nVidia GPU work in 300 milliwatts with their "neuromorphic" accelerators, I have three
im still reading btw just not trying to disturb your
wall lol
Google Coral M.2 TPUs are like $30 on Amazon, etc
It's cheap to add "paradigms" to learn
m2 tpus are that cheap? Whats the use case for them?
Oh, and read Concepts, Techniques, and Models of Computation by Haridi and van Roy, increasingly hard to buy online, and work through it, you will be a god
Any kind of LLM inference you might do on nVidia, but wildly lower-power
Google Gemini runs on the next rev of them
That's why Google can do it so much more-cheaply than OpenAI and Anthropic
capacity of the m2 tpus? like what type of models can they run?
"The Cathedral Book"
I'm going to do a YouTube channel where I work through it all in detail, I have the authors' permission and they gave me all their course notes from later years
wonderful
only the 8 trillion operations per second, dunno if you can get anything done with that :V
Various models have 1GB, 2GB, and 4GB in the "Coral" generation
sounds good enough for realtime vision.
Yeah, though the BrainChip stuff is wildly better-yet
but more money https://shop.brainchipinc.com/products/m-2-card-m-key
not even a heatsink, and you can destroy an nVidia edge inference system with one
You use these with SNNs (Spiking Neural Nets) not Large Language Models
Ideally with event-driven cameras etc not global shutter cameras
Sony, Prophesee, etc
I wouldn't normally link such details in this channel, but IMO this is my "career advice" right now
Don't get stuck following behind the people who already know how to write CUDA and make LLMs, etc, that era is already over for me, get on the next train car
CUDA is a domain specific language for PTX which is a domain specific language for SASS, the latter undocumented by design
You don't have time to learn all three before that era is behind us, IMO
If you absolutely must, learn Triton and its Gluon subset; then your stuff will run on nVidia, AMD, and Intel at least.
But IMO all three of those vendors are uninteresting to me now, and my LinkedIn feed agrees
Cerebras is what I'd buy if you told me to make money running an LLM tomorrow
but my counter-proposal would be to buy an Oxide Computer rack and load it up with BrainChip AKD1500s
why?
180x more power-efficient, way cheaper to acquire
but u cant run LLMs
Anybody who wants to talk more and doesn't feel comfortable doing so here can find me at https://github.com/wilson where I have most things linked up. I am the 174th GitHub user, lol.
How old are you even
LLMs suck, check out TRM (Tiny Recursive Models), Sapient Inc's HRM, and Spiking Neural Networks, etc
47
My buddy Jeremy is GH user 95, but I slept on it, I was into svk at the time and didn't see anything I needed in git, I already had distributed works-on-airplane-before-they-had-WiFi development etc.
But eventually peer pressure made me sign up
To be fair the svk build process was insane, git is way better
svk was like C, perl, and some other crap, all together
(nobody remembers svk)
Oh, and, uhh, 4chan was a replacement for a site I was tired of maintaining, I was the first moderator, sorry about all that
Really?? Thats insane
but I'm part of the reason it inherently had no way to keep logs, I wasn't willing to get involved if there were logs
also I helped with the dynamic IP blacklist, though moot wouldn't let me port us to nginx, freaking Apache :\
I didn't write any of the site code, they wanted to use PHP and I've always refused to use things that state their limitations in the name like that
"personal home page", lol, forget that
#raspberryheaven irc channel, crazy internet dungeon, the source of many things
I'm an operator there :\
I changed my handle to Defiler in 1991 when the Dark Sun AD&D boxed set came out
@pastel aspen Which low level langs is a must to learn?
Zig
Learn to read C but never write it
Ideally also learn to read LLVM bitcode, .ll files
mhm
But don't start there maybe
You should be sorry
I am, yeah
Some things I'm working on are in penance.
I had stepped down from moderating by the time the "Anon" thing happened, I would have come down on that so hard
In retrospect I should have tapped back in and raked them over the coals
moot was 15 when he founded 4chan? Thats so cool
But nobody imagined it would "take off" like that, total memelords etc
Yeah, he's a lot younger than me, I was "Lord of Slots" at the time in our DC++ server
hence them asking me for guidance
I lost the domain when I forgot to renew long ago, but I had a site at hellninjacommando.com/2ch where I had an English front-end translation for the real 2ch, "futaba channel", the thing that 4chan copied
4chan used the same single-letter sections, but changed what the letters stood for to meme things
I was sick of updating the page so I told the channel I was going to stand it down, and moot did 4chan in response
the 4 is an homage to "Yotsuba", a character in a manga by the same author as Azumanga Daioh (hilarious show)
4 is "yotsu" as a counter word in Japanese
"mitsu" in mitsubishi is 3, "yotsubishi" would mean "4 diamonds" instead of the "3 diamonds" the real brand means etc
"futa" is 2, hence futaba channel
So 4chan's name is very much in sync with the ecosystem there
But yeah, never let an "AI" write code for you, or do anything "for" you that you'd ever care to get good at
Just that alone will make you wildly stand out in the modern job market, bringing it back to the topic
You can have it show you some example code to learn a new language etc but don't "use" that, write your own version
If you want a picture, ask it to advise you on how to learn to draw, don't have it generate an image
If you always just stop when you realize you are being "lazy" about something you might actually care about, and try to then do the opposite, you'll never stop improving, and people won't believe your capabilties after not very long
Because nobody actually does that
"I'll start tomorrow"
Sure.
Oh, and try to figure out how to get interested in health and fitness, it's wildly easier to "hack" than people think, though there's a lot to learn along the way, and the benefits are ridiculous
check out "Caloric Restriction with Optimal Nutrition", insane leverage
You never have to "go to the gym" etc, just learn how it all works and you'll know what to do
You can end up looking like a Fight Club extra just with bodyweight exercises and nutrition
I'm 72 inches / 183cm, ~175lbs / 79kg, and I can squat down and flip a 650lb / 295kg earth mover tire with my bare hands, and I haven't been training in a while, been busy with life chaos
and I'm in my late 40s
learn a martial art if you're even remotely motivated by that, Aikijujutsu is my favorite, but there are many good ones, Kenpo, Wing Chun, Brazilian Jujitsu, etc.
Just don't pick one that is a "sport" where the muscle memory isn't actually about combat
It's not about violence, just about not learning bad habits.
Meta-commentary on that, and then I'll get some sleep, it's late as hell.. always be aware of the "muscle memory" you are building.. occasionally ponder if you like where that's going, and redirect if you don't
Starting a new thing always feels hard, just ignore that, never think of yourself as an impostor
Everybody started somewhere, and that's you today, etc.
Most people need 2000 hours of practice to actually get good at something, don't beat yourself up for sucking at hour 1000.
I have a lot of cheats there but they aren't easy to copy, you have to build your own.
(The more things you've taught yourself, the more angles you have to speed up the next thing, more or less, but the "things" in the toolkit are different for everyone.)
Famous naturalist E O Wilson called that idea "consilience", very powerful
My LinkedIn "headline" is Consilience Engineer, so I never forget.
Last thing before I stop tormenting us all for the evening; it may not shock anyone anymore that this is my watch:
Another thing not to forget. Make the most of your days.
Jesus Christ 😆 I wonder what % that puts you in when it comes to adoption... Also am I reading your above message correctly? You were learning about EE at age 6?
Heh, IIRC, it was an Apache vulnerability that led to the site getting compromised a few years back, too. I believe they were still running on a very outdated version
is this meant to be a watch flex or
Read the hands. It isn't an expensive watch.
also please don't include BJJ in this list lol
I sorta don't, but it is easier to find and will teach you some useful things, just definitely not the same things as Daito-ryu Aikijujutsu
How to get money for claude so it can do everything for me
ive done grappling for a while and have competed/"greased" before. anything vaguely related to aikido or aikijutsu is just cringe
Have you tried asking Claude? 😆
you might be right
Cut small expenses temporarily
Skip a few coffees, takeouts, or subscriptions you rarely use
Cancel one streaming service for a month
Set aside a small amount weekly
Claude Pro costs $20/month, so saving ~$5/week gets you there in a month
Use a separate savings jar or sub-account to keep it distinct
Earn a little extra
Sell unused items online
Do a small freelance task, survey, or gig
Use cashback/rewards
Pay for it with a cashback credit card or use accumulated rewards points
Some banks let you round up purchases into savings automatically
Check if you qualify for discounts
Students or educators sometimes get deals on AI tools — worth checking Anthropic's site
If it's for work, ask your employer to cover it as a productivity tool
Once you've saved up, you can subscribe at claude.ai/upgrade. Is there a specific constraint making it tricky — like a tight budget or no card access?
stop
Yeah, don't do any of that.
And don't give money to companies that shred books when scanning them to evade the law.
Not really, my father was an EE student (he already had a Chem Eng degree from earlier), he was pretty new to programming himself, so he got a fellow student to teach me who had been doing it much longer. I just wanted to learn to make games.
my friend said that i need to learn many thing abt Python to create game with godot, can anyone give me some tips to learn GD Scripts?
Hello umm I just started python
And I'm going to start web scraping, is it actually a good idea to start with it
Godot does not use Python at all. GDScript is a completely different language that might have some similarities to Python.
yes i know, but my friend advises that I need to learn Python first because I want to code a discord bot or web-developing too.
Knowing Python is not a prerequisite to learning GDScript as far as I know
If you want to ask about resources for learning Python, #python-discussion is the right channel
ok thanks
My work has drastically changed so much even just in the past 6-12 months alone. i barely write any code, i just ask copilot or claude to write plans, and i edit the plans now -- given i do understand the codebase and business logic pretty well now
even testing is so much faster to write, testcontainers, playwright, etc -- of course not everything 100% will be covered but I can now focus on those other things that can not be caught with written tests alone. i wonder how classic SWE will transform in the next 5 years.
Im hoping to get to a point with claude where i do half an hours work and thats it for the day for me
That would be fun
That would be ideal but some companies like mine (mostly upper management crap) are now expecting much more velocity from devs, not sure how I feel about it. Most of my day is now is just doing manual testing
That sounds to me as multiple concurrent agents
the moment claude cowork becomes available in my team, even manual testing is just going to be a skill 
Hi, what language would you guys recommend to learn?
What kind of developer do you want to be?
I would like to be able to work around the world remotely on the back end, fixing bugs and writing code
I don’t really like web development and all of that
Jobs that are that flexible are pretty rare, and they're usually only for seniors
"back end" is pretty generic. There are a lot of languages that can be used in that space. Python is one of them
Python is the one I heard more but if I’m going to invest my time on learning programming I think I have to learn something good, that’s why I ask for good recommendations
Maybe c++ is better or java script ?
You don't think Python is good?
JavaScript is closely tied to web development, which you said you're not interested in.
It would probably be more worthwhile to learn rust than c++
Oh I don’t mean that python is not good I just want to know if it’s the best one
It's the top ranked language on the TIOBE index.
Alright then I would learn python 🐍
Hi everyone, I am new to coding. I am studying something totally different but now I was thinking to become later on like a sales engineer and I realized that programming can be useful in this. I started learning Python for data in codeacademy, should I also follow the tutorial of Python from zero to hery of udemy?
Oh I also got that course from udemy but I find it kinda boring
Hey, i work a blue collar job and know zero tech people, im in college for computer science right now, I just want to introduce myself, im completely unfamiliar with how these forums work
You're on the right track so far. Every channel has a description you can read to find out what you can talk about there.
What kind of blue collar work are you doing? I was a barista all through uni
youre asking this in the python server
I work at a wood recycling plant. It gets hot and that place is very old school, the most tech thing i can do there is working on bagging machines. I'm doing good in school, BUT I see that learning CS is just part of getting in tech. I'm trying to eventually see how people work as a team building projects and soon maybe get involved in building something when I hit my 2nd year in CS or so.
Honestly there's no need to wait, you can 100% start building small things yourself before second year. In fact, I'd be very surprised if you didn't have to do that as part of your year 1 course.
Yeah, there are a ton of "simulators" you could run for free on even the cheapest computer that can run Discord, and you can learn amazing things without needing your job to help you.
I almost never was taught a single thing on the job in my whole career.
Well before computer science i prompt engineered huge things on my github. But the more i learned about software.. the more i learned what i DON'T know. so i buckled down and decided to go to college. But i was reading that it don't matter if I build personal projects, that employeers want to see i can work as a team. When I'm done with my degree I want to be able to have a solid resume with team projects so I'm competitive. Not just small solo projects I'd be able to prompt with AI. Or, maybe i can show how i made an app with AI then debugged it.
..and I've had admin on a Burroughs/Unisys box, literally "MCP" as in the Tron movie, stack hardware not register hardware, etc. Zero people helped me learn that.
(Amazing CLI actually, ST is the status command, etc, it's even-terser than UNIX, and more-consistent)
I don't know how true what i am saying is but thats why i'm here
There's some truth to it still, but there aren't always great ways to find a "team" as an independent, and personal projects that are "interesting" vs. "fancy" do have an impact, IMO, and a strong one.
"Another Discord bot" is not such a project, etc... and it doesn't have to "work" if it is ambitious.
Visibly vibe-coded things with emoji in the README, very much the inverse signal, I would never show such a thing to a recruiter
Ha, yeah I would highly recommend building things without AI to start. I'd much rather take on a junior with fundamentals than an impressive sounding but entirely unmaintainable project that they don't even really understand.
FYI there are ways to show you can work in a team outside of building group projects, too. Sports, clubs, all sorts. If you really want to work on a larger project, looking into open source contributions is an option. If you find you're regularly using an open source library or tool, it might be worth checking out their GitHub and seeing if they have any open issues you think you could pick up. A fair few projects mark good tickets for beginners.
my cringe classmates put emojis in their readmes before it was cool
Actually a 2018 README with emoji is a flex maybe, yeah... but also cringe-y.
Not to mention describing everything as 'production ready', lol
You can backdate commits on GitHub, don't be fooled 😁
basically every rust repo in 2018-20 had at least three 🚀 emojis
"ready" means you haven't done it yet.. "in production" is different haha
"Fly what you test, test what you fly."
When i was building my project last year with AI i started to think... "How do i know it would work on other operating systems? What if it just says its perfect and it isn't? what happens if the data people entered gets lost? what if it gets hacked? how secure is this thing it says its secure but how do i know? what if an os system gets an update what do i do?
John Arundel, Go/Rust famous guy, just posted this on LinkedIn, and I more or less agree:
How to program:
- Start with the simplest, dumbest idea you can think of. If it works, you're done.
- If not, add the smallest possible complication. GOTO 1.
- There is no step 3.
If you want to be confident in your projects, I'd recommend looking into test driven development 🙂 It's a good way to force you to think through what you're actually building, and to validate the functionality you need is there.
It also ties fairly closely to the view @pastel aspen just outlined, iteratively adding on functionality.
I'll note that when you're starting out, even if you try and be very thoughtful about how you're designing it, you'll likely find it becomes pretty unwieldy, pretty quickly. I wouldn't try and prematurely account for that, but when you feel it happening I'd take it as a signal to do some reading on different design patterns. Writing maintainable code is hard, but we've developed a fair few approaches over the years that can help make it a little less so.
As for the security side of things, I wouldn't worry about that too much to start, but looking into Static Application Security Testing (SAST) and the OWASP top 10 would also go a long way toward helping you keep your code secure 🙂
TDD is rad, BDD is rad, I am working toward moving up to what I call "Semantics-Driven Development" now, where I write the Lean4 proofs first and then the code, etc
Look into the needs-Internet-Archive-now obscure OWASP AppSensor v2 PDF, wonderful idea that basically nobody was ready to actually do, so they stopped telling people about it, but it's very "right", to me.
The thesis is basically that your app is the thing that actually knows if it makes sense for your CFO to be logging in from Thailand at 4am.. Does your org's calendar say he's on a trip there? No? Don't even need to check the password, it's fake. Are they signed into the Thailand VPN endpoint? No? Then for sure it's fake, etc.
Don't use a "firewall", make your app collaborate in making there be context for everything. So powerful.
You don't even really need to read it, the above summary is the main takeaway, the rest is just advice about how to do that.
but it's more fun to figure that out yourself IMO
AppSensor is the fix for the "crunchy on the outside, soft on the inside" security issue that is the bane of all classical approaches; AppSensor isn't fooled by you being on the "LAN"
Just knowing the idea will put you far ahead of the "SecEng" interview pack
Speaking of the 'right' way to do security, I've been working on a tool that lets you determine pre-deployment if a planned IaC change would introduce a privilege escalation pathway.
Graph based tooling exists but is post-deployment, pre-deployment tooling exists but relies on static property checks. I've built something that lets you take the graph based approach, and use it to reason across both the planned changes and the full live environment (including resources not managed by IaC/managed in a different state file), so you get the power of the graph based approach before you actually implement the change. Woo for shifting left
Yessss, do this, very legit idea.
Any IaC change may make the "parameter space" as I think of it larger, which means you need to re-audit your whole test suite.
hey
Also look up "Confused Deputy", get over your shock and horror, and never let that happen.
If possible, arrange for your "root of trust" to live in a safe, not on a computer.
im new to python.should i practice hackarathon problems or i shuld build projects first?
Both
also watch the Help System section here and don't give bad advice, but try to figure out the answer before the helpers chime in
Already built! It works 🙂 I'm building on top of Bloodhound. So I can take a live scan of the environment, parse the planned Terraform change and convert it into a Bloodhound-compatible format, merge those into a composite graph representing the expected future state if the plan were to be deployed, and then use Bloodhound's existing shortest path queries to identify if the planned changes would introduce any new escalation pathways.
Had a brief conversation with one of the Bloodhound co-creators. They think there's tons of potential in the concept, and the only reason SpecterOps haven't pursued it is constrained engineering resources.
Amazing training
do ya'll have githubs or reddits as well? i swear i won't spam you trying to get a job just starting my CS degree but I'd like to keep up with what you guys are doing. thank you for the advice
guys please help me
Be warned, I am the weirdest of the weird, but almost all of my "profiles" are linked here; glad to advise/collaborate/etc in general: https://github.com/wilson
Between the day job and my Master's, I don't really have time for a GitHub at the moment 😁
(I'm working on a GitHub replacement now that they are getting worse, stay tuned. It won't have Git in the name, baking in the VCS tool is a bad idea IMO.)
Hey now, they were briefly back at 3 nines! 89.99 counts, right?
Mine has a 1000x tougher security stance
At the risk of being considered "woo-woo", here's my current list of the "Laws" that apply to our apparent reality:
Where is it?
Nothing deployed yet, but I registered 5 years of VCSector.net
Still nailing down some of the architectural details.
Deployed? Who cares?
gib code
Haha not sharing that until it's shared from the actual platform. Smalltalk-80 blue book style.
Fly what you test, test what you fly.

The code will not be on GitHub
Are you making your own VCS too or just your own forge for Git repos?
Git for now, but I may support jj also, not sure yet; I don't actually love it yet, I tried it last year and it bit me like 10 times in the first three days
I'm down to support other good ones if there are any popular options, hence the name being "VCS" not "Git"
wait a second this is #career-advice
There are lots of "forges" now, some pretty good, but they are all massively complex and insecure compared to what I want. Oh yeah sorry.
But yeah, lemme know if you know of other "Laws of Karma" than I've listed.
My whole career of experiences is modeled 100% by the above though IMO.
@pastel aspen I have a project idea for you. Make a secure sandbox and run malware inside it! Truly puts your skin in danger!!
I've got one actually, eBPF-managed Linux kernel keyring API service that uses Incus to run Podman quadlets, and the quadlets can only access secrets via the keyring service, mlock() in host kernel memory, unswappable, point-of-access auditing
It's not open source, too sharp a knife to share.
You mean its dangerous? how?
Uhh, it's about running malware
That question is why people can't be trusted with it haha
I also don't share my project called Apex Predator :V
whats that
Hope you never find out 🙂
well its not a malware but i can run it safely with ur tool!
Even with my tool, if you let it access the internet, you're done, etc
aw give a hint atleast 😔
💣
"Continuous Schrödinger Bridges, Entropic Optimal Flow, Square-Root Law"
That's all I'll say
I also dont share my project called Cerbereus 😉
I just don't have private repos 🤓
-# I do have some at work
"You are cooked if you run it" thats all il say. And i have never shared it with anyone apart from running it in some vms~
It's good to have some secrets like that IMO, good practice.
its illegal to share 😭
I made it for learning how things like that work
"When guns are outlawed, only outlaws will have guns", etc.
Some finance trading model? Or a smart AI model?
Neither. Even if you guess right, I won't confirm it.
alright we all are allowed to have our secrets :)
Especially me. 🙂
I've overheard quite a few tech people talk at length about restaurant food and motorized travel.
Does it make sense to buy a cookbook and a guidebook so you at least have something to say?
You're overthinking imo, I wouldn't bother learning about hobbies solely in the hopes that it'll help you talk to other techies. I'm sure you can find genuine common ground without trying to manufacture it.
Restaurant food is a genuine interest when hungry...
But what you say is definitely the main strategy to find common ground. I was curious whether a small investment in their small talk made sense.
<@&831776746206265384>
!rule pay
!cleanban 1505240422841192488 Advertising a tech talent agency
:incoming_envelope: :ok_hand: applied ban to @forest pasture permanently.
where do you overhear people talking about food and travel? please dont approach people randomly just to further your network, that's kinda creepy
its really not that hard to mingle with tech people at tech conferences and meetups
can i ask do u guys think bootcamps work as im about taking a software engineering
Even if you learn from doing them, the problem is that employers don't take them seriously
im doing coding on the side as well
In the handful of bootcamp interviews i've sat in, i've noticed that paying for bootcamps is definitely not worth it
so would u say do projects on the side as well so i can build a protfolio
In the only bootcamp success stories I know, the bootcamps were paid for by the people who would employ them after completion
shit so would u say if it was held by a university dont do it
The market is rough at the moment. Rewind a few years and one might have been sufficient to get into an entry level tech position. Nowadays, I'm seeing a lot of grads with internships still struggling.
Not to mention a lot of bootcamps rush you through material so quickly that you don't have an opportunity to actually absorb it properly. It's often just cramming that won't help you develop good fundamentals.
so im done for then cool
do they offer this for free?
no 5k uk to do it with Manchester uni and hyperionDev
Ah, you're in the UK 🙂 It's worth looking into apprenticeships
They're competitive, but a brilliant way to get experience quickly while also earning
i currently work on a atm island and need to learn the basics
my main problem is that im u able to relocate
What's an ATM Island? I'm not familiar
island mull and ATM means at the moment