Finding hard to transition from academia to industry.

[rigid cape]

After graduating in 2021, I continued to work in academia and I haven't had much success transitioning to industry. I have hid a small section of papers I have published. Any advice would be appreciated.

[brave vector]

are u looking for generic SDE positions or what?

[rigid cape]

I am focusing on Machine Learning Engineer or Data Scientist roles, but happy to transition to SWE.

[supple tide]

Resume seems good

Maybe list some databases/ warehouses you used?

Might make it look more industry friendly

[rigid cape]

Only getting interviews for jobs where I would be taking a pay cut..

[calm hemlock]

Yeah that's the market rn. Wait for Jan, more jobs will open up then, but everywhere is offering less for the same jobs than they were 2 years ago.

[rigid cape]

@obsidian veldt

[obsidian veldt]

Are you applying in Europe?

[rigid cape]

Yes.

Well, I apply everywhere. I am based in Europe.

[obsidian veldt]

I'm unfamiliar with European markets and can't give you advice there. From your bullets I'm not sure what you did. For example, what problem did you solve from implementing Cox?

I would assume you know how to process data and apply statistics given your pedigree and experience.

You can remove courses.

Do you have any publications?

[rigid cape]

Cox is a survival analysis technique, we used it to estimate the effect of different covariates on the hazard rate.

[rigid cape]

Yes, 2 conference, hoping to submit 2 more this year.

[rigid cape]

Are there any other bulletpoints that aren't clear?

[obsidian veldt]

Yes but as an example, was it used to find patient survival times?

[obsidian veldt]

Were these in a journal?

[rigid cape]

I guess in machine learning language, the model wasn't used for "inference". We were modelling the data in a statistical sense. The whole study was focused on measuring how the covariates effect the hazard.

[obsidian veldt]

Yes. As an example your MLR position. Developed and applied novel methodologies, setting new benchmarks...

[obsidian veldt]

What was its purpose? Were you trying to solve a problem, benchmark models, messing around?

[rigid cape]

Study how the different features effect financial vulnerability and then reported our findings to the government.

[obsidian veldt]

Great, add that.

Do you see what I'm getting at?

[rigid cape]

But don't I say that in the first bullet point?

Maybe I can add more tools that I used; HPC, packages, etc. to the second bullet point.

[dark lotus]

You’re missing the E in MLE, more DS resume

Did you publish at the last job?

Describe better your PDE research, its motivation, etc

Why would you implement a UNet from scratch in torch? (also its not really an advanced method)

Is the Fourier Neural operator stuff open sourced?

Did you actually implement Cox? Or did you use an implementation? There are good packages for this it is very popular

[rigid cape]

We introduced our own variation of UNet with discretization free properties.

[rigid cape]

Cox had to be implemented from scratch; we were only allowed pre-approved packages because of the highly sensitive data.

But I'll try to rewrite my ML experience.