Design an in-memory file system supporting mkdir, ls, addContentToFile and readContentFromFile (often extended with file versioning), then handle concurrent access.

Model directories/files as a tree or HashMap-of-nodes (a trie-like path map works well); implement each op, then walk through edge cases (path doesn't exist, file vs dir). When the interviewer pivots to concurrency, reason explicitly about read/write locks, granularity, and how versioning changes the locking story. Clean low-level design and data-structure choice under follow-up pressure; the real signal is how you reason about thread-safety and read/write locking at Netflix scale, not just getting the happy path working.

Serialize and deserialize a binary tree.

Pick an encoding (level-order/BFS with null markers, or pre-order DFS), then write the inverse parse cleanly. State the format up front, keep both directions symmetric, and verify the round-trip; aim for an O(n) solution and call out the null-sentinel handling. Ability to convert cleanly between representations, disciplined string/queue handling, and writing testable code in a no-debugger environment (print-statement debugging on CoderPad/CodeSignal).

Merge Intervals — merge all overlapping intervals in a list.

Sort by start, sweep once, extend the current interval while overlapping else push a new one. Explicitly handle touching vs overlapping boundaries; note O(n log n) from the sort. Solid array manipulation, careful edge-case handling on boundaries, and clear complexity reasoning.

Design a system to withstand a DDoS attack (a security-focused design round).

Frame the threat model and attack surface, then defend in layers: anycast/edge absorption, rate limiting and traffic shaping, WAF and bot/anomaly detection, autoscaling and load shedding, graceful degradation of non-critical paths, and observability to detect the attack early. Discuss false-positive cost to real users. Security and resilience thinking under sustained drill-down — Netflix is more likely than other FAANG to dedicate an entire design round to security, spending the whole session drilling into the details. They evaluate availability, blast-radius containment, and detection, not a memorized framework.

Reverse system design: walk me through the largest-scale system you have personally built — its architecture, the hard trade-offs you made, what went wrong, and what you would change.

Pick a real system, lead with the constraints/scale, then narrate concrete decisions, the failures you hit, and the calls that turned out right or wrong. Expect pointed follow-ups (e.g., 'how did you scale it?', 'where does this fall over?'). Netflix explicitly wants someone who has already done the thing, not someone who could theoretically figure it out. They probe depth of real ownership, honest reflection on mistakes, and the ability to defend decisions live.

Design a frequency-capping system for ads: limit how often a given user is shown ads across line items, campaigns, and categories.

Define the capping dimensions and time windows, then design counters keyed per user/campaign/category with fast read-on-serve and eventual durability; discuss approximate vs exact counts at scale, eviction/expiry, and serving-path latency budgets. Cover the data model for advertisers, impressions, and revenue tracking. Real-time counting at scale, data modeling that ties impressions to billing/revenue, and serving-latency vs accuracy trade-offs — a question that surfaces on Netflix's ads-tier teams and mirrors their real ad-serving infra.

Tell me about a time you disagreed with your manager (or a decision you couldn't persuade others on) and how you handled it.

Use a tight STAR story showing candor: you voiced a reasoned dissent directly, sought to understand the other side, and then either changed the decision with data or disagreed-and-committed. End with the outcome and what you learned. Candor and 'disagree and commit' — that you can challenge directly without being combative and operate like a 'fully formed adult' rather than deferring or going passive-aggressive.

What would you do if you realized your manager was about to make a decision that could harm Netflix?

Assume positive intent (your manager may have context you lack), raise the concern privately with specific evidence, and — if you still believe it would materially harm Netflix — escalate to ensure the right people are aware. This is exactly the 'act in Netflix's best interest' / courage value. Ownership and putting company impact above hierarchy or self-preservation; willingness to raise hard issues with evidence — a direct test of the culture memo's values.

Tell me about a time you failed (or received critical/negative feedback) — and what you did afterward.

Pick a real, owned failure, take clear personal responsibility (no blame-shifting), and emphasize the specific behavior change and result that followed. Show you metabolize candid feedback rather than getting defensive. Self-awareness, accountability for both wins and losses, and comfort giving AND receiving direct feedback — central to surviving Netflix's high-candor, Keeper-Test culture.

Netflix Interview Questions

Netflix runs a deliberately decentralized loop — each team picks its own problems, so there is no single standardized question bank — but the shape is consistent: recruiter call, hiring-manager screen, a 45-60 min technical phone screen (1-2 coding problems), then a 4-6 interview onsite spanning coding, a heavily-weighted system design round, and behavioral/culture rounds (including a 'Dream Team' round with a director). Culture fit and system design carry far more decision weight than raw coding, and an offer requires unanimous interviewer agreement.

17 real questions across 3 rounds

Reported via Reported on the LeetCode Netflix company tag (krishnadey30 netflix_alltime.csv frequency data: Time Based KV Store, Merge Intervals, Find Median from Data Stream, and Design Hit Counter all appear), plus candidate write-ups on Glassdoor and 1Point3Acres (including a 'Log Handling System Design' phone-screen thread describing an LRU-like log system with concurrency and deletion). Coding is the least decisive round; problems are often medium-hard 'LeetCode with a practical/systems twist' and team-dependent — some teams skip LeetCode entirely.

6 more coding questions Netflix actually asks, each with how to approach it and exactly what the interviewer is evaluating. Unlock the full set across every round.

Continue with Google

Or run a Netflix mock with the AI interviewer. It asks questions like these, follows up on your answers, and tells you exactly what to fix. The only score that means anything is the one on your real answers.

Continue with Google

Keep prepping

All company interviews