Hi, I am preparing for an onsite interview with FB for a Data Science, analytics role. In one of the interviews, you are given an open-ended question, and your task is to translate it to a structured data question and showcase your analytical¯ approach. A toy example is "Do siblings interact more or less on Facebook?". To solve that, one could think of 1) how to detect siblings, 2) how to quantify interactions 2) how to split the data (e.g., region, age) to provide actionable insight. What else could the interviewer expect? Are there more examples like this that anyone could share? Is there anyone who could evaluate the responses to such questions and highlight things that the interviewers expect that may not be assumed? #facebookdatascientist #facebookinterview #facebookanalytics
All of that to become a Sql monkey? Why?
Do you have personal experience with that? I've reached out to a lot of people and I have gathered a pretty thorough idea of what I can expect. Obviously it depends on your interests. If you want to do a lot of ML then this is not the right place.
But typically the people who self-select into this role can’t do true ML work such as reading a new academic paper and synthesizing those concepts into code. But at least it will give them some exposure to it.
Don’t do it, FB DS Analytics is the biggest scam in the industry
hey, thanks for responding. Why are you saying that?
Nothing wrong with analyst work, but the DS title used to mean what is now Applied Scientist and the like. DS roles that do that work still exist, but this at FB ain't it.
Search SQL questions on Blind for resources beyond leetcode. To practice actual analytics, try case interview prep books. Same idea of breaking down a problem and identifying what a useful result would look like. The example you gave sounds like a stats question. Don't know if that's needed for the role, but unless someone shares otherwise, good idea to review basic stats and hypothesis testing.
The misinformation on Blind is strong. I'm DS FB Analytics - I have barely written 200 lines of sql in the past year and I'm an IC. That said, for the role, think "Data driven decision making". This means you should be able to synthesize situations into data problems and propose a solution. To prepare, just go through a number of situations. Knowing sql alone isn't going to do shit for you. Also make sure you know basic statistics - distributions, testing etc.. Also, contrary to popular belief you don't need to do your interview in sql, you can do it in python or R - people often choose sql because it takes the coding out of it and the focus remains the business understanding of the problems and your data science skills. Don't understand the hatred on blind, but neither Facebook nor people in the role are idiots - and Facebook as a company certainly doesn't pay 400k salaries to people for running sql queries - its all about driving better, more data driven decisions.
I've heard much lower TC numbers. Is it really normal to see 400k for senior ICs in this role?
Yes, I'd say range for senior ics being 325-425 depending on level.
Well again, make sure it's right org, there are some lower paying data science roles here as well. The distinction is whether or not you're part of core xfn team for a product. You are the one setting goals, supporting product strategies etc.. Usually Eng and pm at Facebook will need DS sign-off on goals. If you are not that DS, then that's some other data science role at FB.
My interviewer for this applied data round didn't know what virality means? Should I be worried about my results
That sounds like a fun role what’s your background?
This is the standard FB DS role that many people complain about containing too much SQL :'D. My background is swe, ml and for the past couple of year DS, algorithms and analytics.