Tech IndustryMay 22, 2019
NewFVYY73

Google ML SWE interviews

Has anyone done the ML SWE interview at google that would be willing to share their experience? I have a choice to do onsite as full SWE or as ML SWE which would just swap one of the onsite rounds with an ML round. What is the difficulty of the ML questions/what is the best way to prepare? I have deep applied ML experience from my current job but specifically in the areas we focus on but not sure if it will be enough to cover all my bases in a tougher interview if asked to go in depth on many different topics/very deep in general stats theory.

Add a comment
Amazon bingming May 22, 2019

If you said “some”, chances are it’s not enough. I would suggest doing full SWE.

New
FVYY73 OP May 22, 2019

Some meaning very deep in select areas but otherwise general basics in others - will clarify in post

New
Tc270Nw150 May 22, 2019

If you're good in general system design, it may be safer to go with full SWE.

Glassdoor mdcn May 22, 2019

But you pass the ML round you'll get a better offer

New
FVYY73 OP May 22, 2019

Any idea how much the offer will differ by? Have about 3 years of experience

Facebook Pascality May 22, 2019

This isn't true. SWE and ML SWE bands are the same. Your offer depends mostly on competing offers anyway. It may be true that ML SWE's have a higher chance of getting a competing offer, which will obviously matter.

Google 打打打打打打打打打打 May 22, 2019

[Blind] Check out this post! ML interview questions (Tech Lounge) https://us.teamblind.com/s/qP3yeK4E

New
Tc270Nw150 May 22, 2019

Why is this link not working?

Google 打打打打打打打打打打 May 22, 2019

need to be part of the Tech Lounge

Google frSU26 May 22, 2019

It's riskier I think but still I'd go for the ML interview because then you can get matched with an ML team. A lot of work at Google is dead boring so it's important to make a good team match.

Facebook Pascality May 22, 2019

I got a Google ML SWE offer last year. The ML rounds (I had 2 - might depend on level - mine was L6) are harder to prepare for because they're not standardized. The interviewer might ask anything really. I got asked about transfer learning in one round, and k-means in another. However, ML rounds are more interesting, and less stressful. Fwiw, Google doesn't care what you worked on - unless you select NLP or CV as your specialization. You'll be asked generic ML questions.

HCA Healthcare LXvJ31 May 23, 2019

Are the ML questions (at least the ones you faced) something you might encounter in a rigorous proof-based university ML 101 course or does it go far deeper than that?

Facebook Paillier May 23, 2019

The ML questions are literally whatever the interviewer wants to talk about. In my case, neither were from a 101 class - they were long discussions that meandered.