I heard some people say that none of the algorithms published at conferences like nips/icml/colt etc are actually used in production since they're all impractical for many reasons. I would like to work on algorithms (be it of ML kind or other kinds) at Google and I'm wondering if a PhD in theoretical CS/ML would help me get there.
Depends on what you actually mean by "algorithms". There's 10k+ people working on ads/news feed/search ranking across FANG. If you want to contribute to that, switching teams is probably the best way. So something ML adjacent until you know ML well enough to be a ranking engineer. If your goal is fancy research/novel stuff, a top PhD is a decent choice.
So basically to be what you call a "ranking engineer" is just better to be a regular FAANG SWE and then get to those teams rather than doing a theoretical CS/algo PhD
If fancy novel stuff means being the coolest 'ranking engineer' then yes. If it means being a broke academic then no
My friend once told me that you can work on cool stuff and publish it on Medium. The same boring model will go to production.
We definitely implement “some” of the algorithms/systems published in top conferences
bubble sort
Tech Industry
19h
1692
Why doesn't OpenAI offshore and reduce expense by 80%
Tech Industry
3d
41394
What happens when most of your team is Indian?
Tech Industry
5h
776
Question about women in their 30’s?
Software Engineering Career
18h
555
Do Tech leads make more than Engineering Manager?
Software Engineering Career
12h
2322
L4 Google -> 45 interviews, 5 offers, AMA
Search runs on Bogosort