Meta Data Scientist Interview Experience — SQL and A/B Testing Screen, No Offer

There were 2 SQL questions — one to calculate a total and one to calculate a percentage. Neither question was hard, but given how demanding Meta interviews are, you need to think through a lot of edge cases, or you might be done for.

The case study covered A/B testing and stats estimation. Also not hard, but you had to grasp the key points accurately and quickly. For A/B testing they wanted to test on a random sample, and the issue was how to handle it when the samples can't be independent. Stats estimation was really about not randomly assuming things about the data — you provide a solution based on the data you actually have.

I only managed to answer the above after the interviewer followed up — looking back, I realized not grasping the key point right away was the fatal flaw of this interview. 1-2 days after the interview I couldn't reach the recruiter anymore, so I figured it was dead for good.

Update (2025-02-22): Just got the result — rejected, as expected. For a Meta interview you need to quickly grasp the core of the question; only a conversation centered on the core points is a useful interview. The interviewer was actually pretty nice, and patiently guided me toward the right direction — overall it felt pretty good.