International, MS CS 2024 Grad. Looking for any advice. 100+ Apps not yet shortlisted.

[fresh anvil]

International, MS CS 2024 Grad. Looking for any advice. 100+ Apps not yet shortlisted.

[fresh anvil]

@vocal spoke If you could spend a few of your minutes, I would be grateful. Please take this at your own time and availability.

[vocal spoke]

Sure thing I'll take a look in a bit

[vocal spoke]

research:

• developed advanced methods? OK? what do those methods do? and what led to the 22% number?
• to "stengthen a user browsing internet"? bro what does that even mean??
• you "innovated" scripts? LOOOOL don't use that verb I can't take you seriously
• so how did container orchestration lead to 30% boost? like there's no context that shows the technical detail of what you did
• 45% faster closure? what does that even mean? closure of what??

software eng:

• bro this is one of the funniest bullet points I've ever read LMMAAAOOOOO 🤣 you PENETRATED 15% more users??? LLLOOOOOOOOLLLLL that's way too funny 🤣 don't say you penetrated your users bro that is not a good thing at all. Also I can't connect the work to the metric, which is the main issue here
• in this whole job you kinda just list general responsibilities without really showing what you did

SDE:

• "helmed"? no one says this. "to the publication"? wdym? to the publication of what? what even happened in this bullet point? super awkward very and sentence that doesn't convey any meaning
• OK you made multi threaded code. WHAT did the code do tho? Like technically speaking what did you make that resulted in 30% performance boost???
• can you say how the idle notification helped cut down operational cost? It's not obvious how one leads to the other
• you don't "formulate" a bot, you engineer or create it. Again I don't see how what you did leads to that metric

Your two biggest issues:

1. You say something vague and jump straight into a metric, which makes sound fake or made up. Your bullet points sound like this: Did X which resulted in Y% improvement OK but how did X lead to Y? That's the question I keep having when I read your bullet points
2. I can tell English isn't your first language so it's OK, but some of your verbs are over the top and awkward like "Helmed" and "Innovated"... no one is impressed by these verbs, it's better to use basic ones and focus on the content of your bullet points

And lastly.... you should NOT penetrate your users 🤣 🤣 🤣

[fresh anvil]

Thanks for the in-depth analysis bro.
Here are few things that may have gone unclear:
Research:

1. I guess developed advanced method to analyze Google Drive and GMail kinda answers "What do those methods do? If I'm missing something on this, LMK.
2. Accepted. Will improve this.
3. Thank you.
4. What I meant in this point was, that creating performant microservices, and deploying the service as multi-containers reduced the latency. Thus the performance boost.
5. Correct. It's unclear. Will modify it.

Software Eng:

1. 😂 LOL. I didn't look at this from your POV. But user penetration is a fairly common term "https://www.sortlist.com/blog/penetration-rate/#:~:text=your penetration rate%3A-,Penetration rate %3D (Number%20of%20consumers%20or%20users%20or%20customers%20%2F,by%20that%20product%20or%20service." I will rephrase it for sure.
2. Agreed. Will make this section better.

SDE:

1. Yeah that was me trying to convey "Directed end-to-end development of Hindi and English Automated Speech recognition Models. This work was then published as a paper in IEEE conference.
2. Go it. More technicalities to be added.
3. Not formulated. Got it. Thanks.

The reason behind using those weird verbs was also based on some article I had read that said to avoid re-using same verbs.
But thanks a lot man. That really helped.
I'll try to accommodate all your points and then pitch a new one to you soon.

[vocal spoke]

• you anlayzed gmail for what exactly? you didn't answer the question

• ok so make sure you say you containerized it to improve async or whatever which lead to the performace. you can't just jump to performance boost, you have to say what led to that.

• LOOOOL it's market penetration not user penetration LMAOO

• Wow you got published in IEEE and you didn't even mention that? Clean up that bullet point and make sure to mention that

• You can reuse verbs but not more like twice and three times, so don't worry, it's better than awkward and weird sounding verbs

[fresh anvil]

Well, user penetration is a separate term too 😂 User penetration focuses on a specific user group, while market penetration looks at the broader market as a whole. But yeah this one got to be the most intuitive and riveting review so far. 😂😂
https://konigle.com/glossary/u/user-penetration#:~:text=User penetration in the context,a certain period of time.

[vocal spoke]

HAHAHAHAHA wow such is my ignorance then

sounds good to me 🤣 👌🏼

[vocal spoke]

before I read any further, you're still not saying how you analyzed space in gmail

what are these "advanced techniques"

what's stopping you from describing them?

[fresh anvil]

Part of ongoing paper can't reveal as of now.

As per prof.

[vocal spoke]

but you can be general

like a broad mathematical concept?

anything?

it sounds made up

[fresh anvil]

Yep machine learning and maths

[vocal spoke]

OK

tell me as much as you can without getting in trouble

with kind of neural net?

can you say?

and what math techniques?

can you say?

[fresh anvil]

yeah, sure, The model uses vector embeddings to create a similarity matrix as a starting point and then recommendations are generated by a hybrid system

[vocal spoke]

OK cool

that's really cool

so what are you embedding? duplicate files or?

like what are you targetting

[fresh anvil]

embeddings are generated for every object present in storage that includes the content of the file + the metadata of the file + other features.
These are then stored, and then analysis is run over these.
Targetted objects can be objects that are stale [not accessed in a long time], potential duplicates (yes), and
Then recommending based on user behavior.

[vocal spoke]

OK that's sick

such a shame that not even a little bit is discussed on your resume

[fresh anvil]

It's just that I don't know what to put and I have tons to things to write but only limited space.

*of

[vocal spoke]

Developed machine learning model using PyTorch to index and analyze storage usage statistics on Google Drive and Gmail to target candidate objects for deletion, freeing up to 22% of disk capacity

what do you think?

[fresh anvil]

🤯

[vocal spoke]

i put pytorch as an assumption

you can replace it

[fresh anvil]

That sounds sick.

[vocal spoke]

removed "potentially" sounds mumbo

see how I get to the meat and potatoes of what you did

I zoom in just enough to provide meaningful contenxt to the reader

some would say even more but it depends

I would add some sort of math technique

is there anything you can say?

like a statistical or mathematical technique you used in your analysis?

since it's an academic bullet point

would be nice

[fresh anvil]

Let's say plain cosine similarity for now.

[vocal spoke]

oh that's just the distance of the embedded vectors?

sorry im a noob

[fresh anvil]

Yeah, it's just a distance. But I really can't reveal that technique as it's crux and something that I don't own.

[vocal spoke]

right right

because vector distance is usually handled by the vector DB anyway right?

[fresh anvil]

that's right

[vocal spoke]

Cool that's OK

[fresh anvil]

But what about other points? Do they seem better? FYI I removed PENETRATED

[vocal spoke]

soudns like top secret 🤣

[vocal spoke]

I'll keep reading now

[fresh anvil]

Thanks

[vocal spoke]

i'm just gonna fire off feedback one by one it might be spammy sorry in advance

[fresh anvil]

Shoot me

[vocal spoke]

• how did shared cookies protect users? same problem as above it's kinda too vague and you have to "zoom in"

• 30% due to more parallelism is mumbo jumbo... just say what you containerize and also say what tech you used... you should name drop Docker or whatever you used

• mentoring -> 45% faster shipment is mumbo jumbo.... don't use metrics that sound fake like that unless you're willing to put some effort and actually explain why... if you're too lazy to write and just go "mentoring led to 45% faster deployments" then it sounds kinda fake and made up

I just realized you're doing the "Skills demonstrated" thing instead of highlighting the tech used in bullet point... all good

• why is "PAN" in all caps? is it supposed to mean something?

• "designed end to end developement" doesn't really make sense... do you mean "designed and developed"?

the rest is good

just work on your habit of providing enough context for the bullet point to make sense and your mettric not to sound fake

when you do skills demonstrated, I would just list the tech buzz words, and not soft skills... it makes it crowded

and remove soft skills from the bottom as well it's meaningless

and lastly put a scale for GPA

4.0/4.0

[fresh anvil]

Got it. Thank you so much.
Also, is there any set rule for having at least X % of quantifiable metrics present in the resume?

Cause it's not always possible to quantify every work.

[vocal spoke]

youre right... try your best to quantify otherwise don't

it's ok