Data Science applicants have rated the interview process at Meta with 3 out of 5 (where 5 is the highest level of difficulty) and assessed their interview experience as 100% positive. To compare, the company-average is 59% positive. This is according to Glassdoor user ratings.
Here are the most commonly searched roles for interview reports -
I applied online. The process took 2 weeks. I interviewed at Meta (Menlo Park, CA) in Sep 2017
Interview
First a phone screen with the recruiter who explains the position and the interview process. Then a video interview with a data scientist which consists of a technical coding exercise and some product questions about what could be driving activity.
Interview questions [4]
Question 1
Very disappointing experience from interviewing with FB. Absolute worst and least professional interviewer I have ever had.
* FB does video interviews for the first round (along with coding in an online environment). This is fairly surprising since it can't serve much legitimate purpose. I'm sure their legal team is thrilled at the additional burden they face in discrimination lawsuits since the interviewer clearly can has information about the appearance of the candidate and they can't plead ignorance. Of course perhaps they are looking for unkempt candidates or trying to demonstrate the culture of transparency at Facebook, but read on!
* Interview was 5 minutes late to video interview and had disheveled appearance during whole interview. He was constantly fidgeting and playing with the neighboring chairs and objects on the table and even put his feet on the other chairs and table! An interview for a position is a two-way street, so the interviewer also needs to be on "best behavior".
* In addition to the behavior above which was unprofessional, he interrupted very frequently. This goes beyond steering the conversation. This is really tricky since one person is judging you, nobody wants to say "Please allow me to finish explaining my answer and reasoning" and hope somebody like this won't hold it against you. He possibly spent just as much time talking as I did during the interview answering his questions. The interruptions came both in the coding sections when I explained my thought process before implementing it and during the product questions when explaining possible scenarios. Of course clarifying comments are helpful to make sure the candidate is on the right track, but interrupting somebody just because they want to use a subquery (and you wouldn't) and is going to solve the problem slightly differently than you would is indicative of your lack of emotional intelligence and perhaps also the programming language you're working with.
* He obviously wasn't paying attention during the process and didn't take it seriously. He was constantly working on his computer. Of course he could have been taking notes, but after I answered a question twice and explained my thoughts twice, he stated what the correct answer was--which was in fact what I said twice, but he apparently hadn't been listening then and was occupied with the chair/table objects/computer.
* The interviewer seemed to only accept very specific answers that aligned with his thoughts and preferred answers. This mentality is obviously horrible for open-ended questions like "what could be causing this" or "what data would you look at". You may be looking at this data all the time and know which fields are useful for this task, but when you condescendingly react like the interviewee is a complete idiot for thinking of looking at the timestamp patterns for app usage you don't demonstrate your intelligence but rather your closed mindedness and unfamiliarity and inexperience with working with any type of data except the dataset at your company.
* It is probably clear from reading the interview questions from other candidates and what the recruiters provide you as prep, but this really isn't a data scientist position: it's more or a data analyst or SQL monkey position. For this reason I was borderline on interviewing here anyway but thought, "hey, it is Facebook". I'm sure everybody in the industry knows that this is what FB data scientists do, so while the name cache might carry advantages when looking for a job later, it could actually harm you if you want to be thought of as a data scientist and move on to legit data scientist/software engineer positions.
* The information you receive that you can solve the questions in SQL or Python or R is possibly incorrect. My interviewer wanted it in SQL or Python. Not a big deal since I know all three, but when when this is one of the first things you hear (along with all the rest of the insanity of this interview) it is enough to throw you off your game. SQL seems to be the preferred solution, so I'd recommend forgetting about the others if your SQL is decent.
* The one positive of all of this experience is that the recruiters were professional--both in scheduling the interviews and in explaining the role and answering any questions you may have.
Overall, I agree with some of the other comments here that you might get unluckly with a very egotistical interviewer. Officially I haven't heard back from FB yet, but I would be shocked to advance and really have no interest to work with someone like this. Bad apples exist in all companies, but a bro culture like this really does dampen any interest of anything FB. Yikes.
You have a table with appID, eventID, and timestamp. eventID is either 'click' or 'impression'. Calculate the click through rate. Now do it in for each app.
I applied through college or university. The process took 4 weeks. I interviewed at Meta
Interview
simple greetings, and the inteviewer briefly indroduced himself and what he do in facebook.
Then he just throw out some product-related questions, with a bunch of follow up questions. Remeber to be well prepared for what you are saying, cuz the inteviewer may ask you why for some details, and you need also to make your thoughts well organized, since multiple reasons are required ( I think they expect the insights shown in your answer, not the generalized univeral truth).
after that there is a sql task, not hard, but it seems that not a tiny mistake is allowed. They are not only looking at the correctiness of your answer, but evaluating your tech competence throughout the process. So, be confident. perform like a professionalist.
I applied online. The process took 2 weeks. I interviewed at Meta
Interview
This was an interview for the analyst side of Facebook rather than the ML team. A recruiter reached out and discussed the interview process with me, though I did not advance beyond the phone interview stage. Following the discussion with the recruiter, a phone interview was scheduled with a Facebook data scientist. The person I interviewed with was great to speak with and well-versed in how Facebook works (answered all my questions with clear knowledge of the company).
The interview was conducted via video and a code-sharing website. As with most Facebook interviews, there was a SQL question and a business case question. The SQL question was not as difficult as I was expecting, but I flubbed some basic maths in the process by trying to make it more complicated than it needed to be. I'm not sure if that was the sole reason I did not move forward, but I'm sure there were candidates who answered the question flawlessly.
The business case scenario was interesting, and again I think I did reasonably well but not excellent. There were several terms that I was unfamiliar with (e.g., countermetrics) and I was quite upfront about my lack of knowledge during those parts of the interview. I am coming straight from academia and am not going to fake business knowledge. I prepared quite a bit in learning/brainstorming metrics and business case scenarios, so I'm not sure if more preparation would have helped.
In the end, the process was clear, well-managed, and friendly. Facebook does a great job with this compared to other companies I interviewed with, so I was impressed. I was not particularly interested in the position because the side of Facebook DS I was interviewing for was the analytics side and I am much more interested in ML work - something to be aware of if you are applying.