
Meta Data Scientist Interview Questions
& Process
Real candidates share what happened, how many rounds they had,
and how the experience turned out.
Based on 708 interview experiences · FREE TO READ
Candidate interview experiences
First-hand accounts from people who interviewed at Meta.
Data Scientist
As a candidate, I went through a structured, multi-step interview: first a recruiter screen to check fit and timeline, then a technical interview on SQL, metrics, and experiment design, followed by two virtual interviews focused on product sense and analytical reasoning. The final loop included a deep-dive case and a behavioral interview. Communication was clear, expectations were shared ahead of time, and I got consolidated feedback within two weeks.
- Design an experiment to evaluate the impact of a ranking change on 7-day retention—what would you choose as the main metric, what guardrails would you add, how would you set the sample size, and how do you deal with novelty effects?
Data Scientist
It started with a recruiter call. The recruiter contacted me and asked me to book a time slot for the phone call. In the phone call he asked me 6 MCQ questions, 3 on python and 3 on SQL. I answered all of them correctly. After that I was told to fill out a form with visa related questions like are you authorized to work in USA. After that I was ghosted.
- What is the difference between the max function and the greatest function in SQL?
Data Science
Recruiter screen: Role overview and background fit check. 45-minute conversation with a data science leader: SQL case study with metric design and product analysis, analytical case study based on Meta products. Onsite (5 rounds): One technical case study, two behavioural rounds, two leadership rounds focused on team management.
- Hands-on SQL coding in practice
- Analytical case talk through based on Meta products
Meta Data Scientist Interview Questions
Quoted word for word from Meta interview reports.
“To get the count of users on Facebook between two given dates, how would you write a SQL query, assuming there is a users table with a created_at (or signup_date) field that records when each user signed up?”
Read reports →“Table 1: users_disabled: user_id (int), disabled_reason (string), date (string). Table 2: actors: user_id1 (int), user_id2 (int), probability_of_same_user (double). How do you count the number of users for each disabled_reason in the last 7 days?”
Read reports →“In a binary-class logistic regression setup, suppose one of the coefficient estimates is x and its standard deviation is y. Do you reject the null hypothesis?”
Read reports →“You have two tables, A with date and userid and a second table with date and search_id. For the last 7 days, how many users performed 10 or more searches?”
Read reports →“Using SQL, how would you get a distribution of the rolling 7-day average amount of money spent per person, split into purchase categories?”
Read report →“The notification product is going to launch a new feature, which is a new notification type: when your friends attend an event, you will receive a notification. How would you measure whether this new feature is successful?”
Read report →“Write an SQL query to find the users who logged in during last week. What KPI do you think would fit best for this task?”
Read report →“The events team wants to send notifications to users whenever a friend is attending an event in the same city where the user lives. How would you assess if this is a good change to introduce?”
Read report →“Case Analysis 2 - How would you measure the impact of using a smaller font in News Feed for text, video, and photo posts?”
Read report →Formats, difficulty and experience
Across all 708 Meta interview reports.