Facebook has two data science teams, the "core" team (which is the one featured in various news stories), and the product/growth team. I interviewed for the product side, so a core data science interview might be substantially different. From my experience, I get the sense that "data scientist" on the product side is a misnomer, and they are more of the business analysts; while the "core" team is similar to other data science teams in the bay area. The first interview is a phone screen (or on-site if you're local). They asked a few SQL and probability questions, and it was over in ~30 minutes. My interviewer was friendly and enthusiastic, and the questions weren't particularly difficult. The second interview was on-site, and involved talking with 5 different people (or groups of people). Each interviewer had a specific task they were asking about, including business analytics/product health questions, stats/probability theory, algorithms, and more SQL. The questions weren't particularly hard or out of left field but they're looking for you to answer confidently and efficiently. The people were overwhelmingly friendly and tended to be enthusiastic about working for Facebook (although, at least 2 of my interviewers subtly admitted they were casually looking for jobs - so, I suspect even employees at Facebook aren't 100% fulfilled by their positions).