Meta interview questions for software engineers

Meta's loop is the fastest among FAANG and the most numbers-driven. The coding round is timed tightly — two medium problems in roughly 35 minutes — and the behavioral round is built around the question "what did you ship, and what changed because of it?". This guide covers the process, the question patterns, and the cultural cues that decide an offer based on public Glassdoor reports and Meta's own published engineering ladders.

The Meta interview process

Meta runs a tight loop. Recruiter screen (30 minutes). Phone screen (45 minutes, one or two coding problems on coderpad). Onsite (full day, virtual or in-person, five rounds: two coding "Ninja" rounds, one system design or product architecture round, one behavioral "Jedi" round, one optional domain-specific round). Debrief happens within a week, then offer. The total timeline is typically 4-6 weeks — fast enough that you should not start prep the week before recruiter contact.

One quirk: Meta levels you after debrief, not before. You apply to a general SWE role and the committee decides whether to offer you E4, E5, or E6 based on signal. This means under-performing on system design at E5 expectations gets you an E4 offer instead of a rejection — a softer landing than Google's binary outcome, but a signal that your scope evidence wasn't strong enough.

Top 10 technical questions to prepare

Meta coding rounds prize speed and clean code under time pressure. Pattern recognition matters more than exotic algorithms. These are the recurring shapes.

1. Subarray sum equals K — prefix sums + hashmap. Hint: a single pass is enough; rehearse this until you can write it in 5 minutes.
2. Validate binary search tree — DFS with (min, max) bounds. Hint: in-order traversal also works but bounds are cleaner under pressure.
3. Merge intervals / insert interval — sort by start, sweep. Hint: practice the in-place variant — interviewers ask "what if input is sorted?" mid-problem.
4. Number of islands and variants — BFS or DFS on a grid. Hint: rehearse the version where you must count distinct island shapes — pattern hashing trips people up.
5. Lowest common ancestor (with parent pointers and without) — recursion or two-pointer climb. Hint: both versions show up in the same loop; practice both.
6. Random pick weighted — prefix sums + binary search. Hint: the binary search edge cases (bisect_left vs bisect_right) trip up senior candidates.
7. Trapping rain water — two-pointer or monotonic stack. Hint: have one approach you can deliver in five minutes flat.
8. K closest points to origin — heap or quickselect. Hint: heap is the safe answer; quickselect is the impressive answer; know both.
9. Word break and variants — DP with memoization. Hint: the variant where you must return all sentences requires backtracking — different code shape entirely.
10. Serialize and deserialize a binary tree — BFS or DFS encoding. Hint: pick one encoding and rehearse both directions; switching mid-solve loses points.

Top 5 system design topics

1. News feed — Meta's flagship system design question. Practice fanout strategies, ranking, freshness, and caching.
2. Real-time messaging (Messenger or WhatsApp) — delivery guarantees, presence, end-to-end encryption tradeoffs, group fanout.
3. Live video streaming — encoding pipelines, CDN, adaptive bitrate, latency floor.
4. Notification delivery system — fan-out at scale, push vs pull, rate limiting per user, deduplication.
5. Distributed counter — Like counter for a viral post, sharding strategies, consistency vs availability tradeoffs.

For each, lead with capacity estimation. Meta loves order-of-magnitude reasoning — interviewers will push back if you skip the "this is roughly 5 billion QPS" step. Know the numbers for Meta's published scale: 3+ billion DAUs across the family of apps, hundreds of billions of messages per day.

Top 5 behavioral questions (Jedi round)

1. Tell me about your most impactful project in the last year. Lead with the metric, then the story.
2. Describe a conflict with a peer or manager and how it resolved. Specifics about the disagreement, not just the outcome.
3. Tell me about a time you took initiative on something outside your scope. Bias for action is core to Meta's culture.
4. Walk me through a technical decision you made that turned out wrong. The reflection matters more than the mistake.
5. How do you decide what to work on when priorities conflict? Mention impact estimation, stakeholder alignment, and willingness to push back.

Tips specific to Meta's culture

"Move Fast" is not just a slogan; it shapes the interview. Hesitation in the Ninja round reads as inability to ship. Practice talking through your approach in under 60 seconds before coding — that crisp opener buys you trust and gives the interviewer the signal they need before timing pressure kicks in.

The behavioral round uses the "What I Get Done" framework from internal performance reviews. Every story needs a measurable outcome: revenue moved, latency reduced, users affected, code shipped. "We improved the team's velocity" without a number scores at the bottom of the rubric. "We cut p99 latency from 480ms to 110ms across 12 services, which unblocked a feature shipping to 200M users" scores at the top. The numbers don't need to be huge — they need to be specific.

Senior candidates: have at least one story where you pushed back on a leader's decision and won. Meta promotes engineers who change minds, not engineers who execute orders. This is the single highest-leverage prep item for E5+ behavioral rounds.

Frequently asked questions