How to come back to academic machine learning career after absence due to health issues

Question

Back in 2018 I started a master's degree in CS focusing on ML with the goal of eventually obtaining a PhD. This did not work out for two reasons: 1) most of the ML faculty left for another university. 2) I was diagnosed with depression and could barely keep up with classwork let alone do meaningful research.

Since then I have graduated and worked in non-ML-related generic software engineering jobs, always holding on to dear life due to my mental health issues and being an average performer at best. Thankfully, this year I have been approved for a new form of therapy which has resulted in great improvements. I feel almost entirely normal again and my productivity has skyrocketed. I am able to read and implement ML papers in my free time again.

I now feel determined to go back to my original plan. But the obvious problem is that my CV post Bachelor looks entirely unimpressive and going back to do yet another MS in CS seems like it would raise red flags. At this point I am not sure if there is a realistic way forward into a PhD program. Getting ML industry experience? Trying to self-publish? Going for a non-CS MS degree (e.g. stats)? I am also unsure of whether I have to/should explain my situation in my CV or motivation letters.

Essentially I am unsure how admission committees evaluate non-traditional candidates with obvious gaps in their resumes and whether a comebeback is a) realistic (if so, what paths have you seen such students take?) and b) has to be explained.

Answer 1

I also took an extended mental health detour before getting to my PhD, and my industry programming experience has been very exciting to a lot of the academics I work with because competent programmers are surprisingly rare in computer science. (It makes sense if you think about it, computer science researchers don't spend much of their time on programming.)

When applying for a PhD, the admission committee is trying to decide if you're likely to successfully complete the PhD. You need to present your experience in a way that will make them confident in you as a candidate. If your master's degree transcript is unimpressive, you may want to acknowledge that this was due to a health problem which has now been resolved. I would suggest that your CV should cover all the relevant experience that makes you a good candidate, and your motivation letter should tell the story of why you want to do the PhD and how the experience in your CV sets you up well for this.

If you are struggling to be accepted to PhD programmes, one way you can make yourself more attractive as a candidate is participating in relevant contests/challenges, preferably ones which involve writing a paper about your submission. You want something that involves researchers but is open to anyone. It doesn't matter if you don't score highly (although a high score is nice!), the point is to show a basic level of competence. As an example of the kind of thing I mean, there's an annual challenge run by PhysioNet with an associated conference which has detailed instructions on preparing a paper as the assumption is that not all teams are within academia.

Answer 2

Here are some of the tips that could strengthen your profile and eventually might make ready to get back on track.

1. You can do some ML courses and upload the certificates of completion on your LinkedIn profile.There are so many courses available on platforms like coursera, udemy, udacity, and so on. Pick the one that suits you best.
2. LinkedIn is a nice platform and one of the must-haves these days to get connected with like minded people. You can join certain groups/communities where people look for other ML folks without even needing a fancy ML degree.
3. Start exploring Kaggle competitions and submit your entries without even worrying whether it would lead to the top of the leaderboard.
4. If you are interested in gaining some experience, you can try freelancing.
5. You can always attend workshops, and academic conferences as a visitor to learn what people are doing research on and engage yourself in discussions with real professionals. You can also avail the opportunity to address your concerns and get even more practical and sensible advice.

Good Luck!