Data scientist analytics at FB

Interesting. Didn’t get the impression that I would be building ml models in the position seemed more like a statistician role. But also didn’t get the impression it was about etl/sql either. As an aside regarding your point though think in the next 5 years there’s going to be very little opportunity to do anything in ml as a traditional ds. Because most ml models don’t perform nearly well enough to justify the cost associated with hiring people to build them and common tasks are being automated well enough that you just need someone who can understand what model to apply and how to manipulate the data. From my perspective if one wants to go into ml the only longer term career options are heavy into engineering or heavy into basic research.

It’s not just FB, it’s many tech companies, you have to clarify when you interview that a/b testing and sql is of no interest to you.

Would rather do that than the useless projects that 90% of MS DS works on, MSR is the only worthwhile DS at MS

At alt_qq, is this really true? Ron Kohavi's group seems to be at the vanguard of online experimentation, and many other MS DS roles have an ML element to it. I've been considering moving to a place which has more non-eng ML (which is what I previously did), so would like to know if MS DS is indeed useless.

You could perhaps blind more. This impression is pretty standard for FB. I have personally known ppl who boast about ML/ Data Science but get paid / work as data analysts. But like any other company, FB would have ML/ data science roles where the titles would do justice to the kind of work they do. Knowing the TC would be the best way to differentiate if ppl are actually working as data scientists ;)

I think SQL monkeys isn't an accurate characterization. A better way to think about it is in terms of being a management consultant that takes disparate pieces of information, constructs a narrative with recommendations, and leaves execution to others. SQL is the primary tool used to extract data that creates that narrative, but it's not thoughtless work. This describes 80% of FB DS. Others veer towards more technical work (I've done so, and do use mainly SQL and R, but I also work on graph embeddings, use optimization techniques for decision making, use a fair bit of sampling theory). I'd say that the added technical rigor doesn't help me with my rating (a narrative may be more useful for the team), but it keeps me sane.