Amazon interview questions for software engineers

Amazon is the only major tech company where the behavioral round determines the offer more than the coding round. The 16 Leadership Principles run through every interview, the bar raiser holds veto power, and the rubric explicitly weights ownership, customer obsession, and frugality. A candidate with strong code but weak LP stories regularly gets down-leveled or rejected. This guide breaks down the process, the question patterns, and the LP framework using Amazon's public principles and public Glassdoor reports.

The Amazon interview process

Amazon SDE loops follow a recognizable shape. Recruiter screen (30 minutes). Online assessment (90-120 minutes of timed coding plus a work simulation for some teams). Phone screen (60 minutes, one coding problem plus LP behavioral). Onsite loop (5 rounds, one of which is the bar raiser): two coding rounds, one system design at SDE-II+, one hiring manager round, one bar raiser. Total timeline: 4-8 weeks depending on scheduling.

The bar raiser is the load-bearing piece. They're a trained interviewer pulled in from outside your hiring team, specifically to push back on "good for this team, not good for Amazon" decisions. Their round is structured around Leadership Principles with relentless follow-up. Expect "tell me more about that", "what would you do differently", and "what was the outcome" probes until they hit bedrock. Vague answers get exposed within 30 seconds.

Top 10 technical questions to prepare

Amazon coding questions concentrate on graphs, trees, dynamic programming, and OOP design. These are the patterns from public reports.

1. Number of islands and its variants (max area, distinct shapes). Hint: master DFS and BFS, then practice the variant where you must avoid revisiting via a seen set.
2. Course schedule / topological sort — detect cycles, return order. Hint: Kahn's algorithm is the easiest to articulate aloud.
3. Word ladder — BFS on a word graph. Hint: bidirectional BFS shows seniority; mention it even if you don't implement it.
4. Lowest common ancestor — recursion variant. Hint: clarify whether the nodes are guaranteed in the tree before coding.
5. Merge K sorted lists — heap or pairwise merge. Hint: both work; explain the complexity difference between them.
6. OOD design: parking lot, elevator, vending machine — class hierarchy and state machine. Hint: drive from the user stories, not from inheritance diagrams.
7. Find median from data stream — two heaps. Hint: rehearse the rebalancing rules until automatic.
8. Maximum subarray (Kadane) and the variant on a 2D grid. Hint: the 2D variant is the differentiator at SDE-II+.
9. Coin change / minimum coin problems — DP on subset sums. Hint: clarify whether each coin is unlimited or single-use before writing recurrence.
10. Tree traversal with custom serialization — DFS preorder with null markers. Hint: same encoding for serialize and deserialize cuts code in half.

Top 5 system design topics

1. Distributed cache — sharding, replication, hot keys, cache stampedes.
2. Job scheduler at scale — durable queue, worker pool, retries, dead letters, monitoring.
3. Real-time bidding / ad serving — low-latency lookup, budget enforcement, request-response under millisecond SLAs.
4. E-commerce checkout flow — inventory locks, payment integration, idempotency, eventually-consistent inventory.
5. S3-like blob storage — consistency model, replication, lifecycle policies, large-object support.

Amazon system design rounds love operational concerns: monitoring, alerting, oncall runbooks. Sprinkle these naturally — "we'd export a per-shard latency histogram and alert if p99 crosses 200ms" lands higher than a pure architecture diagram.

Top 5 behavioral questions (Leadership Principles)

1. Tell me about a time you took on something significant outside your area of responsibility. (Ownership)
2. Describe a situation where you disagreed with your manager — what happened? (Have Backbone; Disagree and Commit)
3. Tell me about the most innovative solution you've delivered. (Invent and Simplify)
4. Describe a time you had to dive deep into data to find the root cause of a problem. (Dive Deep)
5. Tell me about a time you had to deliver more with less. (Frugality)

Use STAR — Situation, Task, Action, Result. Every story needs a measurable result. "We shipped it" without a metric scores at the bottom. "We shipped, reduced p99 latency 40%, and the feature is still running 18 months later" scores at the top.

Tips specific to Amazon's culture

The 16 Leadership Principles are not posters — they're the rubric. Print them out, study them, and pre-write a 90-second STAR story for each. Aim for 12-16 distinct stories covering all 16 principles with at least one story applicable to each. The interviewer literally has a checklist; you want at least one strong match per principle they probe.

Use "I" not "we". Amazon's culture explicitly demands individual ownership signal. "We built this thing" is the most common downgrade trigger in behavioral rounds. Even when work was collaborative, frame your role: "I owned the API design, recruited two contributors, and wrote the migration plan." This isn't bragging — it's giving the interviewer something to grade.

One non-obvious principle: "Are Right, A Lot" implicitly demands a story where you changed your mind based on new data. Senior candidates who can't name a time they were wrong score weakly here. Prepare a clean "I was wrong about X, the data showed Y, here's what I shipped instead" story.

Frequently asked questions