Meta Data Science Interview 2026 - Learn from My Mistakes

0:00 One year ago when I came to US I knew nothing about interview process to the big tech companies. So when recruiter reached out to me on LinkedIn from Meta asking if I was open to interview for a data science position in the office in Menllo Park Meta HQ. I was like beyond happy like are you kidding? Yes I'm open to interview. Of course this was my shot my moment and I blew it.

0:31 Of course, when I didn't get the job, I was like devastated. But then I reflected on it and I thought there are lessons I learned along the way. And this video is made for me in the past for for you now where I break down 200 hours of my preparation. I show you all the materials that I used, all the topics that were covered. I break down step by step what happened throughout interview process. So hopefully you or maybe me in the future can get that data science job. My name is Andre. I'm a data scientist in the US. I also built my own product in AI and on this channel I talked about career productivity startups and just join me in this journey. First you have a call by recruiter to get noticed by a metal recruiter or talent sorcerer. You have to have a strong either strong CV and referrals or option B what I had is a strong LinkedIn profile with lots of connections with a huge visibility. You can look at my profile now and what I have and as well as my current role which is at the company called Brilliant and due specifically to the fact that I knew how to get LinkedIn to that point uh met recruiter found me and then the first step is actually a recruiter call which I knew nothing about what's going to happen. I knew nothing about the role. I was just like coming in blind but it's actually quite easy. Mostly what meta recruiters looking for in data scientist is your experience with experimentation. The more the better and handson experience I was asked about and I was drilled about the specific roles that I have in designing experiments and getting them from start to finish. The people that get rejected at this point are the people that don't really have experience in uh getting the experiments from start to finish. So you have to be deep into stats and you have to like show that you know what you're talking about because working at meta even from job description like part of what you're going to do is you're going to evaluate and launch and analyze lots and lots of experiments just because for data science in product as per job description you're going to get yourself in lots and lots of experiments. You're going to launch them with product teams.

3:08 You're going to get them from start to finish. You're going to analyze them. So, you have to have hands-on knowledge in that area. I passed the call and I got invited into technical screen. And then recruiter mentioned something really interesting. He said that data scientists from Meta organize prep session for every candidate that's now in this data science interview loop. And I was invited as one of the candidates. uh would you like to join it? And it was a zoom meeting organized where data scientists from Meta talked about the business cases like how they go on to solve product cases, what you have to focus on, how you have to communicate.

3:49 And I came to this call prepared like I'm going to learn something today. Oh, it's going to be fun. And then can you imagine my face when I joined this zoom call for this prep session and then 300 more data scientists join as well and I'm like oh this is quite a competition that I'm in like this is a lot of people that want to get the same job that I want that are probably wellqualified for a lot of them it's probably not their first rodeo and I got a little bit of fear right then and there. Although the prep session was really good. So they talked about how you talk through the cases uh what is important and what not. The most important part of that for me was when you are talking through the product case you have to always think about the platform as a whole. like Zeta as a company there's there are many different products like Instagram, WhatsApp, Facebook and they're all kind of together as an ecosystem and when you have an experiment in one part of that for example experiment in feed it affects WhatsApp maybe it affects Instagram when you think about the feature that you want to implement in one of the products it might be cannibalizing getting some attention from another feature. So it's kind of widen your field of view a bit.

5:20 That was really important for me there. And as I was getting into technical screen, I was told that technical screen is going to be mostly light product case is going to be 40 minutes and then light SQL question. I use SQL like for entirety of my career. I've been in data science. I've been in data work for like seven years already. It has been it's a long time and I use SQL almost daily but still there are some resources that I use to kind of get knowledge sharp because you have to be perfect in these kinds of screens. So I used data leour I used spread a scratch I use interview query I used lead code I like polished and polished and polished my skills. The most important part here that I learned from other mock interviews that I saw is it's not really important how you solve the question. It's more about communicating thorly from start to finish because as I learned from recruiter like the biggest red flag is actually just going straight to coding.

6:31 Even if you are 100% perfect, but you jump straight to coding, they fail you right on the spot because it's like red flag. You have to communicate first because their approach is all the technical ability, all of the hard skills are easy if you're coachable. They do not know what's going on inside your head. So, you have to let them know what you're thinking at any part of the technical script. The day of the technical scream comes and I'm like really nervous. And you have to understand that at this point like I'm a seasoned professional. I have seven years of interviews. But this is meta we're talking about, right? This is a big league. This is like a lot of money and a lot of prestige and whatnot and a great products and like this is like you know the company. And at that point when the day of the technical screen comes I'm really nervous right I'm just it's this is the big shot for me like this kind of money that you can get uh in this type of role this kind of exposure opportunity like everything around it is just like I don't know it was one of the mistakes that I did in this mental preparation is I put all of this in this pedestal like you know really high I like I got to get this job. I really have to get this job. And when you have that attitude towards some opportunity like some interview or anything else it screws with your head like it really messes up the way you think because you have to have emotional detachment from the process. like all the other interviews throughout my career where I have been really successful. I was treating it like it was another day. It it wouldn't matter.

8:27 I could get the job or maybe I wouldn't get the job but like life moves on. But here it was so important to me that kind of on the day I was sweaty. I'm spaghetti. I was shaking. And then all the preparation that I had, it like kicked in like this and I was like, "Oh, okay." So I got the SQL question, I got the product case and I just I don't remember how I solved them. I remember that after the technical screen, I was looking like this. I was pretty shook. I was 100% sure that I didn't get it. So the first important lesson here is mental preparation is gold. Like you have to have cold nerves when you going into that because like that is 80% of the success. And then the second lesson came like two days after when recruiter called me and said it was perfect perfect like down to the last word. You got everything right like only positive feedback, no negative feedback, nothing to work on. Just keep up what you're doing and for the onsite you're going to be great. And I was like I was so sure that I didn't get it. And then when you are that nervous, it plays tricks on your mind. It makes you think that you are worse than you actually are. So I double down before the onsite on like mental preparation. Like I was talking to myself like this is just another job.

9:53 This is another opportunity. And when the on-site day came, I was much more prepared. So what happens on on-site? I got sent very extensive guide from recruiter. It was like one of the best guides that I had that described really well what's going to happen and what kind of questions they're going to ask and how I'm going to be evaluated and so on and so on. So there were four sections to on-site. First is technical question the same as SQL. The most tricky part was not like getting the specific code in place but rather how to decipher question. So it was more focused on how can you choose the right metrics like we want to evaluate for example is if this feature is good or bad and then you have to come up with metrics based on your the data set that you get. how would you define success of that feature and then code those metrics. So from coding perspective nothing too hard maybe occasional window function or two joints group buys but then this part is uh actually kind of evaluates your business sense and domain knowledge and stuff like that. So once again strata scratch interview query uh data leour all of these resources I'm going to drop them in the comments check them out. Second one was behavioral and basically in behavioral part of meta interview it doesn't matter whether it's software engineer job or a data science job any other job mostly you're going to be evaluated on meta's core qualities like what they value is like think long term product first move fast and so on you can find that on meta website and it's just like the question are kind of aligned hint uh on whether you have those qualities or not. So if you have solid star structure but when you're thinking of star focus more on action and result like a lot of the people they just go on and on and on about situation and task and I worked with coaches and mentors that pointed out to me that the most important part is action and result. action and result. You set the context a bit and you have to set it good. But then talk about what you do and what you get. And if it's quantifiable, like if it's in numbers, that's perfect. If it's not a numbers then uh just get some kind of reference for the interviewer to understand like what actually happened and then spend a little bit of time of talking what happened after some situation or problem or task because the long-term thinking is like really important for any kind of role and specifically for this role like not only what you did in the moment to put out the fire but also like what you did so there won't be such fires in the future. Right? Then there was analytical execution which is more how you have hands-on analytical knowledge. Here you're going to be evaluated. There are going to be some stats question. You have to know like law of large numbers, some different distributions like binomial. Uh central limit theorem is going to come up for sure like averages, medians, small modes, nothing too crazy. Uh there could be and I got one of the cases on machine learning. But if you know like regressions, linear logistic regression, if you know some classification models, how to evaluate them, you're going to be fine. It's mostly about when you have a problem, some sort of data problem, like you need to evaluate something or make decisions with data, how you approach that decision. One of the types of questions that can come up in this type of interview is some metric is down by 10%.

14:07 What are you going to do to evaluate this drop? And then you are going to use a framework that I tell you is you have to come up with hypothesis. And this is one of the parts that you're going to be evaluated on and scored on on your interview. And these hypothesis should be mutually exclusive and collectively exhaustive. Like when you're investigate some kind of data problem, you have to really structure your answer. So you have hypothesis that cover different areas and are complete together and do not really intersect. So you could go in one direction or another direction, but what they're looking for is how well can you define the entirety of all of the reasons that might happen to some metric or another. You have to have domain knowledge for sure. like you have to know the products of Meta Instagram, what what they do, what the kind of features they have, but it's more about how you apply your data knowledge. And here one of the things that they look for is cross functional collaboration is how well you always interact with managers, engineers, and how you do not work in isolation like you work with others and you know when to work with others and when to work alone etc. And the last part is analytical reasoning which is focused on some kind of product opportunity and then experimentation and it basically evaluates like what they really value is do you have a product sense how you use that product sense and how you use that product sense in context of experimentation so it's mostly going to be that so I prepare extensively I got a coach on prepfully I had several coach sessions s when I went through mock interviews, I uh prepped on data interview which was great because it helped me to get into some of the more difficult cases and then I had some materials to prep for machine learning and I'm going to drop the list of materials of course below the video.

16:20 Make sure to check it out. And actually now I even got look at that. I even got book for illustrated guide to neural networks and AI. Uh this is not an ad. It's just like one of the prep that I might have for future interviews. This is an awesome channel start quest. I use them for prep all the time. So I just had kind of all these resources. I put them together. I put in the hours to actually get it right. I prepared mentally and then the day of on-site happens right four interviews in one day it's tough like it's endurance game right it's a marathon and then the first interview technical skills nailed it behavioral interview went just great like I prepared it was something that I could prepare in advance analytical execution about I got some ML question and I got some probability questions went great and then the last interview comes analytical reasoning and I'm like I've done so many of these product cases. I just know all kinds of metrics that you could get. I know how to set up an experiments. What can you do to me that I haven't seen? And then the product case that I get is just like in the domain in the part of the meta ecosystem that I really haven't thought about at all. And it was just like completely different parts that I prepared and I'm like, geez, okay, okay, I I can do this.

17:53 And then I completely blew it because I was overly prepared on some of the more mainstream meta products like I was uh doing a lot of cases on Instagram, on WhatsApp, and then on some parts of meta ecosystem, I wasn't prepared well enough. I'm like, okay, shocks. I immediately knew that I blew it. And then after three weeks of waiting, I got unfortunately you were not selected, but like let's stay in touch. And I'm like, all right, okay, what are you going to do? And at that point, I thought that I only get one chance to get on the meta.

18:31 I'm like, shucks, I didn't get it. And then I found out that you can actually get on your second, third, fourth try. like it doesn't really matter that you failed once like they will still consider you after a year or like half a year. I'm like wow so maybe in the future I'm going to get to the interview again and like I already prepped for it. I already have all the materials. Maybe in the future I'm going to say to them uh unfortunately I did not select you as a company I would like to work for. I wish you all the best in looking for a right candidate. Maybe I'm going to be the one that says it to them. I don't know. I think that this experience have been incredibly helpful to me to identify some of my weaknesses and actually to upgrade myself as a data scientist. Partly to that experience, I got my current job at another company as a data scientist. And I don't know, it's all life. It's all good. And if you are on this journey, I wish you only the best. And on this channel, I'm going to talk about career productivity. I'm going to talk about some of my projects that I'm building uh as a founder. And I'm going to just share with you techie entrepreneurial career way uh that I'm on. And you're welcome to join me on this journey. Good luck. I wish you only the best. Bye.