Transition from Physics PhD
Graduating now with a Physics PhD from Harvard doing mostly experimental work, recently squeezed in some device simulations and modeling. The only CS class in my entire life I have taken is called Data Science which I found myself pretty comfortable in. Spent the past 6 years on coding side projects: webapps, IoT devices and backend, blockchain apps and an internship at a tiny quant fund. So, 0.5 YoE I guess. I am sick of academic lab research and want to move into a coding/fintech job. Started leetcoding on Easys after an online DS&A crash course. Did some ML online courses and tried applying them to my PhD research. I'm familia with Python, Node, SQL, Mongodb, numpy-pandas-mpl, docker, React. 1. Should I do DS or SWE? I prefer WLB over TC. TC should ideally be over 100k. 2. How/What are the chances of leveraging my PhD so I did not waste the prime years of my life on this paper? What level will I be entering at? Looking in the Boston area, maybe NY.
Depending on YOE but you will likely enter as L5. Earning the same as someone without PhD. Paper doesn’t matter in FAANG.
Masters and PhDs start at L4 at Facebook and Google.
Google L4 equals Amazon L5, and this dude above you is from Amazon. levels.fyi
You will be a great candidate for DS. DS loves PhDs. Better wlb usually. Not sure about Boston salaries but should be similar to NYC, I’d aim for 150 TC, minimum (you could prob get above 200 easily).
Data Scientist is the better fit. You would leverage your PhD, have better WLB (probably), an easier time interviewing, and probably start with higher TC (given your background). Anchor point should be 120k base in Seattle, higher in SF (maybe 150ish? Not sure tho). You shouldn't accept lower than 100k base anywhere.
How much Leetcode should I prepare for DS interview? Is solving Mediums within 30 mins sufficient, or do they rarely throw Leetcode style questions?
In my experience FANG won't hire data scientist straight out of a PhD. Most data science roles require industry experience. And if you're interviewing outside FANG for DS leetcode won't help much at all.
Depends on role type/company. Startups usually don't have much LC but more domain related. FAANG is heavy LC
Math PhD from a top 10 institution, got into Google on second try. Data science isn't enough to get into a FANG. You need to learn data structures and algorithms. Languages don't matter at all for entry level interviews. Pick one of (python, Java, c++), find a coding interview book and supplement that with an algorithms text (for reference), a book in your language (or audit a course), and plenty of leetcode medium. It will be easier to get into an ML or applied ML role, partly because you'll understand advanced math concepts better than your average code monkey and partly because those orgs still place value on original PhD research. But that doesn't matter if you can't pass the interview loop. Build your Harvard physics network as much as you can while you're still there. That's the main thing your PhD work will have going for you, and its value may not become clear for a few years but there is value in it for networking, job referrals, etc a few years down the road. More of them than you think will end up in industry in three years. Make friends with CS people if you can, that will build your network further.
Discard coding, obtain soft skills, sell life insurance by pounding your credentials. Also why is everyone answering so confidently on DS career track? Are you all actually DS?
Apply at Amazon as L5 Research Scientist. If you’re very comfortable with Leetcode mediums then you can try L5 Applied Scientist, which pays a little more than RS but carries the risk of getting assigned mundane software work which might bore you and result in losing your edge! Also take the Amazon MLU course for prep. And sorry, those positions are not around 100k. You need to be open to earn 180k-250k!
I hope that things are going well for you. I am curious about what you are doing nowadays
Let's just say your chances of leveraging your PhD so you did not waste the prime years of your life on your paper is the only thing lower than your chances of getting into FANG.
Love your positivity!
Friend had a PhD in Physics. Thesis on Quantum Computing or something like that. 1st job out, worked in the Intel Fab for a few years and said fuck this. Studied DS in the meantime. Jumped to a SW startup doing absolutely nothing related to his PhD (just data science stuff). Year or so later, learned enough SWE, Algos, and systems design and is a vanilla L4 SWE at Google now. So yeah, that's a lot of years to devote to a Physics PhD and an erstwhile stint in semiconductors to start out as an L4 SWE.