Meta Interview Questions — Page 133

Meta·Machine Learning Engineer·Onsite - System Design / Architecture

May 2026

Meta ML engineer interview focused entirely on designing a place recommendation system for a maps product. One long open-ended question that branched into a dozen sub-topics. Dense and a bit exhausting, but not unfair.

Design a machine learning system that recommends places a user might want to visit, like restaurants, parks, or local attractions, for a maps or local discovery product.
How would you define training labels and the prediction target for this recommendation system?
What data sources and features would you use, and how would you engineer them for this system?
Walk me through how you'd handle cold start for both new users and new places.
How would you address risks like popularity bias, feedback loops, spam, and fairness across different types of places and regions?
How would you set up online serving for this system given real-time context like current location and time of day, and what latency constraints would you design around?
How would you run experiments and manage feedback loops in a live recommendation system like this?

“This is one of those questions that feels manageable until you realize how many directions it can go.”

View Post

Meta·Data Scientist·Technical Phone Screen

May 2026

Meta DS interview with a pretty involved SQL problem covering date arithmetic and user segmentation. Two-part question, more depth than I expected for a technical screen.

Given a users table with signup dates and an activity table with daily presence records, write SQL that labels each (user_id, activity_date) row as 'new' if the activity date falls within 30 days of signup (inclusive), and 'old' otherwise. The output should cover a specific date range and avoid multiple labels per user-day.
Using the labels from part one, compute two aggregates over a rolling 30-day window ending today: the count of active new vs old users per day, and a single summary row with total distinct users by label across the whole window. Also explain your assumptions about timezone handling and whether bounds are inclusive or exclusive.

“Took me a minute to internalize that 'first 30 days inclusive' means signup_date + 29, not + 30.”

View Post

Meta·Data Scientist·Technical Phone Screen

May 2026

A product analytics case for Meta DS, structured around a hypothetical e-commerce expansion into a new product category. The whole thing was basically a guided business case with four distinct sub-tasks, and it went deeper than I expected on the math and experiment design side.

Given an e-commerce platform with 8M monthly active buyers, estimate the monthly impacted population, incremental orders, revenue, and gross profit from adding a new Home Office product line, including an attach effect from existing purchases.
Compute the break-even timeline for a $2M fixed setup cost plus variable costs, and show how sensitive that break-even is to plus or minus 20% errors in AOV and conversion uplift.
Identify the three most critical assumptions in this business case and describe how you would validate each one using pre-experiment analyses like log data, surveys, or cohort analysis, before committing engineering resources.
If the pre-test business case clears the hurdle, how would you design the initial experiment, including targeting, success metrics, guardrail metrics, and how you'd ramp exposure?

“The setup hands you all the numbers so you'd think it's easy, but I kept second-guessing the attach piece.”

View Post

Meta·Software Engineer·Onsite - System Design / Architecture

May 2026

Meta SWE onsite, object-oriented design round. The problem was a cloud file system with users, quotas, and a bunch of methods to implement. More moving parts than I expected for an OOD question and it kept growing.

Design an in-memory cloud file system class that supports files with unique names and sizes, users with storage quotas, and an admin user with infinite capacity. Implement methods to add users, add files (by admin or by a specific user), retrieve file sizes, and delete files.
Add a method to retrieve the N largest files whose names start with a given prefix, sorted by size descending and then name ascending for ties, formatted as 'name(size)'.
Add a merge_user method that transfers all files from one user to another, combines their remaining capacities, and deletes the source user.
Implement backup_user and restore_user methods: backup snapshots a user's current files, and restore re-applies that snapshot while skipping any filenames already taken by other users.

“The core CRUD stuff was fine.”

View Post

Meta·Technical Program Manager·Technical Phone Screen

May 2026

TPM loop at Meta, one question about analytics on a product I'd already discussed earlier in the conversation. Short and focused, but the follow-up nature of it made me realize I hadn't set up my earlier answer very well.

Walk me through the analytics you used for the product you described earlier.

“I fumbled this a bit because I'd already talked about the product but hadn't really touched metrics the first time around.”

View Post

Meta·Machine Learning Engineer·Technical Phone Screen

May 2026

Meta MLE coding round, one question the whole time. It was algorithmic, not ML-specific at all, which threw me a bit. The edge cases are where it gets painful.

Implement next permutation in-place: rearrange an array into its lexicographically next greater permutation, or reset it to ascending order if no greater permutation exists. Constant extra memory only.

“I knew the algorithm in theory but fumbled the edge cases badly.”

View Post

Meta·Product Manager·Onsite - Product Sense / Strategy

May 2026

Meta PM interview, one question about monetizing Messenger. Not much context to go on but it's a classic product strategy case.

How would you monetize Facebook Messenger?

“The obvious answer is ads and I went there first, which felt a bit lazy in retrospect.”

View Post

Meta·Machine Learning Engineer·Onsite - Behavioral / Leadership

May 2026

Went through a behavioral round for an MLE role at Meta. Three questions, all pretty standard soft-skills territory, nothing technical. Left feeling okay about it but not great.

Tell me about a conflict you had with a coworker. What was the situation, how did you handle it, and what came out of it? And what would you change if you had to do it again?
Walk me through the project you're most proud of. What was the context, what did you specifically do, what was the measurable impact, and what did you take away from it?
Why are you leaving your current company, and why does this role specifically appeal to you?

“The 'what would you do differently' part is where I fumbled.”

View Post

Meta·Software Engineer·Technical Phone Screen

May 2026

Did a Meta coding round that was pretty standard algorithmic stuff. One question on missing integers from a sorted array, nothing too wild, but the binary search angle is easy to miss if you haven't seen it before.

Given a sorted array of positive integers and an integer k, find the k-th missing positive integer not present in the array.

“Started with the linear scan and it worked, but they pushed me on efficiency.”

View Post

Meta·Software Engineer·Onsite - System Design / Architecture

May 2026

Meta system design round, one big question about building a code judging platform from scratch. Pretty dense scope for a single session but they seemed more interested in how you structured the tradeoffs than whether you hit every component.

Design a system like LeetCode: users can browse problems, submit code, get it judged in a sandbox against test cases, and see a verdict along with runtime and memory stats. Consider problem storage, the submission pipeline, sandboxed multi-language execution with resource limits, queuing, results storage, and optional leaderboard or contest features.

“This one sprawled fast.”

View Post