Meta's interview process in 2026

Meta's hiring process is fast-paced and structured. For engineering roles, expect a recruiter screen, two coding interviews, a system design round (E5+), and a behavioural round. Product and data roles follow a similar structure with role-specific variations.

What's changed recently: Meta now places more emphasis on practical coding (real-world problems over puzzle-style questions), and system design interviews increasingly test your understanding of AI/ML infrastructure — reflecting Meta's strategic pivot toward AI.

Meta interviews are known for being time-constrained. You typically get 35–40 minutes for a coding round, and interviewers expect you to code quickly and correctly on a whiteboard or shared editor. Practice speed as much as accuracy.

Coding questions (Software Engineering)

1. Given a binary tree, return all paths that sum to a target value.

Difficulty: Medium | Topics: Trees, DFS, backtracking

Meta loves tree problems. Use DFS with path tracking. Key insight: paths can start from any node, not just the root.

2. Implement a function to validate a valid parentheses string with wildcards.

Difficulty: Medium | Topics: Greedy, stacks

The wildcard (*) can be (, ), or empty. Use a two-pass approach (left-to-right and right-to-left) or track min/max open count.

3. Find the kth largest element in an unsorted array.

Difficulty: Medium | Topics: Quick-select, heaps

The quick-select approach gives O(n) average time. Discuss the trade-off with a min-heap approach (O(n log k)) for streaming data.

4. Serialize and deserialize a binary tree.

Difficulty: Hard | Topics: Trees, BFS/DFS, string processing

Use pre-order traversal with null markers. This is a Meta classic — they want clean, production-quality code.

5. Design a data structure that supports insert, delete, and getRandom in O(1).

Difficulty: Medium | Topics: Hash maps, arrays

Combine a hash map (value → index) with a dynamic array. On delete, swap the target with the last element.

6. Given an array of meeting intervals, find the minimum number of conference rooms required.

Difficulty: Medium | Topics: Sorting, heaps

Sort by start time, use a min-heap to track end times. The heap size at any point is the number of rooms needed.

7. Longest substring without repeating characters.

Difficulty: Medium | Topics: Sliding window, hash sets

Classic sliding window problem. Meta uses this as a warm-up. They expect you to solve it in under 10 minutes and then discuss optimisations.

8. Clone a graph.

Difficulty: Medium | Topics: BFS/DFS, hash maps

Use BFS or DFS with a visited map that maps original nodes to clones. A good test of your ability to handle references and cycles.

System design questions (E5+)

9. Design Facebook News Feed.

Key topics: Fan-out on write vs. fan-out on read, ranking algorithms, caching layers, real-time updates.

What Meta wants: A deep understanding of the trade-offs between push and pull models. Discuss how celebrity accounts (millions of followers) change the architecture. Mention ML-based ranking.

10. Design Instagram Stories.

Key topics: Media upload/processing, CDN distribution, ephemeral storage (24-hour TTL), viewer tracking, sequential playback.

What Meta wants: How you handle media at scale — upload reliability, transcoding pipelines, global distribution, and the 24-hour expiration lifecycle.

11. Design a real-time chat system (Messenger).

Key topics: WebSockets, message ordering, delivery guarantees (sent/delivered/read), group chats, offline message queue, end-to-end encryption.

What Meta wants: Trade-offs between consistency and availability. How do you handle message ordering in a distributed system? What happens when a user comes back online?

12. Design a content moderation system.

Key topics: ML classification pipeline, human review queues, appeal workflows, false positive handling, real-time vs. batch processing.

What Meta wants: This is increasingly important at Meta. Discuss the trade-off between speed (catching harmful content fast) and accuracy (avoiding false positives that impact user trust).

13. Design a recommendation engine for Facebook Groups.

Key topics: Collaborative filtering, content-based filtering, graph-based recommendations, cold start problem, diversity vs. relevance.

What Meta wants: How you combine multiple signals (social graph, engagement history, content similarity) into a unified ranking. Discuss how you avoid filter bubbles.

Behavioural questions

14. Tell me about a time you had to ship something with a tight deadline.

What they want: Speed and pragmatism. Meta's culture values shipping fast. Show that you can cut scope intelligently without sacrificing quality.

15. Describe a conflict with a teammate. How did you resolve it?

What they want: Emotional intelligence and directness. Meta values open, candid communication. Don't describe a conflict you avoided — describe one you addressed head-on.

16. Tell me about a time you received critical feedback. What did you do with it?

What they want: Growth mindset and self-awareness. Show that you actively sought feedback, processed it constructively, and made concrete changes.

17. What's the most impactful project you've worked on?

What they want: Scale of impact and your specific contribution. Quantify the impact (users affected, revenue generated, performance improved) and be clear about your role.

18. Tell me about a time you made a mistake in production.

What they want: Ownership and systematic thinking. Describe the incident, how you responded, and — crucially — what you changed to prevent it from happening again (process, tooling, testing).

Product management questions

19. How would you improve Facebook Marketplace?

Framework: Segment users (buyers vs. sellers, categories), identify the biggest friction points, propose 2–3 improvements, define metrics, and prioritise by impact.

Meta context: Marketplace competes with eBay, Craigslist, and specialised platforms. Think about trust (scam prevention), discovery (search/recommendations), and transaction completion.

20. You notice that daily active users for a feature dropped 15%. Walk me through your investigation.

Framework: Verify the data (instrumentation change?), segment (platform, geo, user cohort), check for external factors (seasonality, competitor launch), look at the funnel (where are users dropping off?), form hypotheses, and propose quick tests.

21. Should Meta launch a dating feature in a new country?

Framework: Market analysis (TAM, competition, cultural factors), strategic fit (does it help Meta's mission?), localisation requirements, regulatory considerations, launch strategy (phased rollout vs. big bang).

Data science questions

22. How would you design an A/B test for a News Feed ranking change?

Key topics: Metric selection (engagement, satisfaction, time spent), experiment unit (user-level randomisation), duration, network effects, long-term vs. short-term metrics.

Meta twist: News Feed experiments have network effects — a change for one user affects what content their friends see. Discuss how you handle this.

23. How would you detect fake accounts at scale?

Key topics: Feature engineering (account age, activity patterns, social graph analysis), graph-based detection, adversarial robustness, precision/recall trade-offs.

24. Explain how you would build a model to predict user churn.

Key topics: Feature engineering (login frequency, content creation, social connections), model selection, survival analysis, cohort analysis, actionable predictions.

25. A model's precision improved but recall dropped after retraining. What happened and what would you do?

Key topics: Threshold analysis, data drift, class distribution changes, feature importance shifts, model diagnostics.

Meta-specific preparation tips

Move fast

Meta's culture is built on speed. In your answers, emphasise how you shipped quickly, iterated based on data, and avoided over-engineering. "Move fast" doesn't mean "move recklessly" — it means reducing the time between idea and validated learning.

Think in billions

Meta operates at a scale that most companies never reach. When answering system design questions, always consider: what happens at 3 billion users? What happens when this runs across 20 data centres? Scale is not an afterthought — it's the first thought.

Code cleanly and quickly

Meta coding interviews are time-constrained. Practice writing correct, readable code under a timer. Aim to solve medium-difficulty problems in 20 minutes, leaving time for discussion and optimisation.

Practice with realistic mock interviews

Reading questions is not the same as answering them under pressure. Try a mock interview calibrated to Meta's interview style to practice coding under time pressure and get instant feedback.

---

Preparing for Meta? Start a mock interview tailored to Meta's question style, or use company intel to research Meta's latest interview patterns and culture signals.