Meta Interview Questions — Page 146

Meta·Product Manager·Onsite - System Design / Architecture

Jun 2026

Did a product architecture round at Meta for a PM role. One question, pretty technical for a PM interview, left me second-guessing how deep they actually wanted me to go.

Walk me through the technical architecture of your product.

“I went broad first and that was probably a mistake.”

View Post

Meta·Software Engineer·Onsite - System Design / Architecture

Jun 2026

Meta SWE system design round focused on building a copyright violation detection system for uploaded images. The interviewer pushed hard on adversarial robustness and kept challenging a two-model design, which made the second half of the session pretty uncomfortable.

Design a system that detects whether a user-uploaded image infringes on any image in a copyrighted-image registry.
How would you fine-tune the vision-language model for this copyright detection task?
What if the user uploads a collage image where only one section infringes on a copyrighted work?
How would your system handle adversarial text or watermarks overlaid on an image to evade detection?
If someone takes a phone photo of a copyrighted physical print and uploads it, does your system treat that as infringement?

“I went with a VLM embedding pipeline into an ANN index, which felt like the obvious path.”

View Post

Meta·Software Engineer·Technical Phone Screen

Jun 2026

Meta coding screen for a software engineer role, basically a maze problem dressed up in a few different ways. Nothing too wild but the follow-up questions on scaling pushed me more than I expected.

Given a 2D grid maze with walls, open cells, a start, and a target, determine whether the target is reachable from the start and return the shortest path using BFS.
How would you handle edge cases like the start equaling the target, a fully blocked maze, or when the start or target cell itself is a wall?
If the maze has weighted edges instead of uniform movement costs, how would you adapt your approach?
How would you handle very large mazes where memory becomes a bottleneck? Walk through the trade-offs between A* with a heuristic and bidirectional BFS.

“I jumped straight to BFS which was right, but I fumbled the neighbor enumeration early on.”

View Post

Meta·Software Engineer·Onsite - Coding / Algorithms

Jun 2026

Meta SWE coding round, got a BFS graph problem where the starter code was deliberately broken and you had to fix it rather than write from scratch. Interesting twist on the usual format.

You're given a grid with a start cell and an end cell. A BFS implementation is provided but it's buggy because it never tracks visited nodes, so it keeps re-enqueuing cells and either loops forever or times out. Fix the bug and return the shortest path length in steps, or -1 if the end is unreachable.

“The fix itself isn't hard once you see it.”

View Post

Meta·Software Engineer·Technical Phone Screen

Jun 2026

Meta software engineer interview with two design-focused coding questions. Both leaned on data structure fundamentals, prefix sums and stack design, nothing flashy but you had to actually know your stuff.

Design a class that supports efficient range sum queries on an immutable integer array, where the query function can be called many times.
Design a max stack that supports push, pop, top, and also peekMax and popMax operations in addition to standard stack behavior.

“Prefix sum is pretty textbook but the part that tripped me up was thinking about initialization cost vs query cost.”

View Post

Meta·Data Scientist·Technical Phone Screen

Jun 2026

Brutal technical screen for a DS role at Meta. The whole thing was essentially one massive experiment design problem that kept branching into harder sub-questions. Felt like a grad school oral exam more than a job interview.

Design an A/B test for a search ranking change in a two-sided marketplace where sellers can affect each other and consumers influence each other socially. Should you use cluster randomization (by seller or geography) or ego-cluster randomization? Define your exposure and exclusion criteria precisely.
Given a mean cluster size of 50 and an intra-cluster correlation of 0.02, calculate the design effect and how much larger your sample needs to be compared to individual randomization.
How would you prevent treatment contamination when reassigning clusters, and how would you validate the experiment using invariant metrics and a sample ratio mismatch check under clustering?
If cluster membership changes over time, how would you design a rehashing schedule and what analysis approach would you use, specifically around mixed-effects models with cluster random intercepts?

“This is where I spent most of my time and also where I stumbled first.”

View Post

Meta·Data Scientist·Technical Phone Screen

Jun 2026

A technical screen for a DS role at Meta that went deep on experiment design. The whole session was basically one long power analysis problem broken into pieces, and it got harder as it went.

Given a baseline friend-accept rate of 9% and a target of detecting a 0.5 percentage point absolute lift, calculate the per-arm sample size and how many calendar days the experiment would need to run, assuming a two-sided test, 80% power, equal allocation, and 10 million eligible users per day.
How does your sample size estimate change if the true standard deviation turns out to be 20% higher than what you assumed?
For a three-arm experiment with a shared control group, where you want to detect a 0.3 percentage point lift for either treatment versus control at a familywise error rate of 5%, which multiple comparison correction would you use and why? Then recalculate per-arm sample size and total duration with 20% traffic to control and 40% each to the two treatment arms.
What are the consequences of checking your experiment's p-value every day against a naive p < 0.05 threshold, and how would a sequential testing approach like O'Brien-Fleming change your stopping boundaries?
When would you switch from a standard variance reduction approach to cluster-robust standard errors, for example clustering at the feed level rather than the user level, and how does that affect your required sample size?

“I knew the formula but fumbled the variance calculation at first.”

View Post

Meta·Data Scientist·Technical Phone Screen

Jun 2026

Meta DS interview with a meaty product ML question that covered basically everything at once. It was one of those sessions where you realize mid-answer that you've bitten off more than you can chew.

What business goals and engagement metrics would you track for a restaurant recommendation feature inside a social app?
Which behavioral, demographic, and social features would you include in a restaurant recommendation model, and why?
Which model would you choose for this recommendation problem, and why might logistic regression be a reasonable starting point?
How would you evaluate whether your logistic regression model is performing well? Define precision, recall, and accuracy, give their formulas, and explain which metric you'd prioritize for this use case.

“I went straight to click-through rate and kind of stalled there.”

View Post

Meta·Software Engineer·Technical Phone Screen

Jun 2026

Meta data engineering interview focused entirely on a product analytics case around the private accounts feature. Five-part question covering everything from feature framing to experiment design. Dense and exhausting but actually pretty interesting if you like this kind of work.

Define the private account feature and explain the value it creates for users and the platform, including the key trade-offs involved.
What entities, dimensions, and facts would you model to analyze this feature? How would you handle a user's privacy state changing over time?
Propose north-star and guardrail metrics for the private accounts feature, specifically ones that would reveal engagement differences between private and public account users.
Engagement for private-account users has dropped week over week. Walk through your investigation plan, including the slices, funnels, cohorts, and control groups you'd examine.
Recommend one or two experiments or product changes to improve the private accounts feature. Define success metrics, guardrails, and expected trade-offs for each.

“Felt like a warmup but they actually pushed back a bit when I glossed over the trade-offs.”

View Post

Meta·Software Engineer·Technical Phone Screen

Jun 2026

Meta software engineering interview with two coding questions back to back. Both were tree and linked list problems, nothing too exotic on the surface, but the follow-up constraints added real pressure.

Given a binary tree where each node holds an integer, and a range [low, high], write a function in C++ that returns the sum of all node values that fall within that range. Walk through your approach, complexity, and explain how your solution would differ if the tree were guaranteed to be a BST.
You're given two unsorted singly linked lists of integers. Merge them into one sorted singly linked list in nondecreasing order, reusing the existing nodes. Implement in C++, explain the algorithm and complexity, and describe a linear-time approach if both input lists are already sorted.

“Started with a plain recursive DFS and it clicked pretty fast.”

View Post