1 year working as a DS / QA after PhD in Stat. Don't see much impact in my work, and tired of doing analysis. Thinking about switching to ML but haven't made decision as it sounds a big career reset. Love coding, have ML-stat research experience, and happy to give half year or more time to leetcoding if needed. Options considering: a) internal switch to other DS team that has more ML related work b) internal ladder transfer to SWE, then find ML role; c) interview with other companies directly for ML roles d) switch to other companies where DS has more opportunities to productionize ML models. Questions: 1) for a), any team come to your mind? 2) for b - d, which one is more feasible? Or no difference at all as all require same-level leetcode practice? Blind seems suggest leetcode solely is enough for SWE interviews. Do I need study system design, etc.? 3) for d), any companies to suggest? TC: 220K, L4 Struggling to make decision, appreciate much any insight!
Is it Quantitative Analyst position at Google?
Yes, L4. Made edits
I am in same boat and tried (and failed) c several times. To be fair I haven't leetcoded much. I often passed the ml/stats screen but failed technical phone screen which is usually like for swe roles. Effort wise d) or a) is easier but matured companies with production ML systems tend to seggregate ML and DS and often there is no overlap. For d) startups are a better bet. If you have time to leetcode c) is the most reliable track to ensure ML work.
Which kind of phone screen questions failed you? SWE coding or other? In your case, was leetcode the main / only hurdle or also need prepare other stuff? Thx!
Algo questions like you see on leetcode. Another hurdle was not having prior production ML experience. Smaller companies often needed that prior experience. I learnt a bit of spark/hadoop and tried to demonstrate that I could deploy and scale my model. Some we're convinced,some we're not.
Which teams? Curious to know
Medical Brain does a lot of ML. Parts of Geo do more interesting modeling and some ML rather than just analysis. It's not a big career reset if you're done significant coding somewhere. People that have both skill sets tend to jump back and forth when the need arises. Would seem silly to leave Google without trying another team or two, or a ladder transfer.
Good points. Thanks!
I think C/D are great options. There's no harm in trying a couple of phone screens - worst case, you get an idea of where to improve plus make a connection for the future when you're prepared. Yeah, most companies will want to see a good grasp of production-level programming. But even if you don't have the skills right now, many companies will love to see your enthusiasm/potential if you're spending time on LeetCode and/or GitHub. And if you're looking to transition, I'm thinking the earlier, the better. It may be more challenging to find a role compensation-wise if you've been locked into DS work. You'll have that big Google pay package and the disparity between your/their expectations could grow since you'd need to take a step back given less of an ML background. Hope that makes sense!
Thanks much! do make sense. Any advice around SWE interviews? Leetcode is enough or need learn other CS fundamentals?
Leetcode is great, CS fundamentals even better. You'll ultimately have a strong chance if you're open to a (relatively) junior ML Engineer role. Sure, it doesn't feel great to take the 1-year step back, but it'll put you ahead of the pack given your professional experience is somewhat relevant. And set your TC expectations appropriately if possible!