Amazon Backend Engineer Interview Questions (2026)

1. 1Online Assessment (SDE OA)60 minMedium

   Two timed coding problems plus a workplace-simulation and logic section; the main gate for freshers and India volume hiring.

2. 2Phone screen45 minMedium

   One coding problem plus 1-2 Leadership Principle STAR questions with an SDE.

3. 3Coding loop round60 minMedium

   DSA problem to working code, followed by assigned-LP behavioral questions in STAR format.

4. 4System design loop round60 minHard

   Design an Amazon-scale service with capacity math, plus LPs; low-level/OOD design substitutes for junior candidates.

5. 5Hiring Manager round45 minMedium

   Team fit, project deep dives, and Deliver Results/Bias for Action stories with the manager you would report to.

6. 6Bar Raiser60 minHard

   An interviewer from outside the team stress-tests LP stories and overall bar with the hardest cross-examination of the loop; holds veto.

1. Q1

   Design with ownership boundaries: Design a reusable API client for Alexa with auth refresh and request signing

   MediumLow Level DesignAPI client with auth

   How to answer:

   Separate HttpClient, AuthProvider, TokenStore, RequestSigner, RetryPolicy, and Serializer; refresh tokens safely and avoid logging secrets. Make ownership, idempotency, and failure modes explicit.

2. Q2

   Write production-quality, testable code to solve this: Write code to call multiple Kindle services concurrently and return partial results safely

   HardCodingAsync fanout

   How to answer:

   Use async tasks/futures with per-call timeouts, collect successes and errors separately, limit concurrency, and preserve result ordering if required. Add boundary tests and robust error handling.

3. Q3

   Design with ownership boundaries: Design a low-level audit logger for Prime Video administrative actions

   HardLow Level DesignAudit log component

   How to answer:

   Model AuditEvent, Actor, Target, Metadata, Sink, and Redactor; ensure append-only semantics and safe handling of PII. Make ownership, idempotency, and failure modes explicit.

4. Q4

   Write production-quality, testable code to solve this: Generate all valid configurations for a small Alexa rule set with constraints

   HardDSABacktracking combinations

   How to answer:

   Use DFS/backtracking, pruning invalid partial states early. Complexity is exponential; explain pruning and output-size limits. Add boundary tests and robust error handling.

5. Q5

   For Amazon's fulfillment routing flow, write production-quality backend code to implement stale-while-revalidate with request coalescing while meeting limited memory per worker. Describe the main function or class interface, the core data structures, and the edge cases you would test

   MediumCodingCaching

   How to answer:

   A strong answer proposes a small, testable interface and uses serve fresh or bounded-stale values and allow one refresh per key. It handles retries, invalid input, timeouts, and cleanup explicitly, and states O(1) cache lookup with bounded refresh fanout.

6. Q6

   In a Amazon backend interview, solve this DSA problem from Kindle reading progress: identify hot keys from a high-volume request stream. Provide the algorithm, prove correctness, and analyze complexity under 10k requests per second

   HardData Structures & AlgorithmsCaching

   How to answer:

   Model the problem with count-min sketch with heap validation or exact counts for bounded keys. The target solution should achieve sublinear memory for approximate tracking, handle empty inputs and ties, and explain why simpler brute-force approaches do not scale.

7. Q7

   In a Amazon backend interview, solve this DSA problem from payments reconciliation: choose objects for eviction under size and value constraints. Provide the algorithm, prove correctness, and analyze complexity under 100k events per second

   HardData Structures & AlgorithmsCaching

   How to answer:

   Model the problem with knapsack approximation or greedy by value/size depending on constraints. The target solution should achieve O(nW) exact pseudo-polynomial or O(n log n) greedy, handle empty inputs and ties, and explain why simpler brute-force approaches do not scale.

8. Q8

   Design cache-aware read APIs with explicit cache-control semantics and invalidation hooks for Amazon's Kindle reading progress read/write workflow. Specify endpoints or RPCs, request and response schemas, error handling, authentication, idempotency, and pagination where relevant

   MediumAPI DesignCaching

   How to answer:

   A strong answer defines resource-oriented REST endpoints or clear RPC methods for create, retrieve, update, list, and audit operations. It includes stable identifiers, authorization boundaries, validation rules, explicit error codes, rate limits, and backward-compatible versioning.

9. Q9

   Design a highly available regional service for Amazon's payments reconciliation using multi-layer cache hierarchies, invalidation fanout, hot-key handling, and stale reads. The system must handle 100k events per second, zero-downtime deployment, and clock skew. Explain architecture, data flow, consistency, and operations

   HardDistributed SystemsCaching

   How to answer:

   A strong answer decomposes the system into stateless frontends, partitioned services, durable storage, and async processing. It justifies partitioning, replication, leader election or quorum decisions, and regional failover, uses idempotency and back-pressure, and defines SLOs, alerts, and rollback mechanisms.

10. Q10

    Design a eventually consistent workflow for Amazon's fulfillment routing using multi-layer cache hierarchies, invalidation fanout, hot-key handling, and stale reads. The system must handle 50 million daily active users, limited memory per worker, and a thundering herd. Explain architecture, data flow, consistency, and operations

    HardDistributed SystemsCaching

    How to answer:

    A strong answer decomposes the system into stateless frontends, partitioned services, durable storage, and async processing. It justifies idempotent commands, sagas, retries, compensation, and reconciliation, uses idempotency and back-pressure, and defines SLOs, alerts, and rollback mechanisms.

11. Q11

    Design a global low-latency platform for Amazon's Alexa intent history using multi-layer cache hierarchies, invalidation fanout, hot-key handling, and stale reads. The system must handle 1 billion stored records, a 150 ms p99 target, and duplicate client retries. Explain architecture, data flow, consistency, and operations

    ExpertDistributed SystemsCaching

    How to answer:

    A strong answer decomposes the system into stateless frontends, partitioned services, durable storage, and async processing. It justifies multi-region routing, data locality, cache hierarchy, and conflict resolution, uses idempotency and back-pressure, and defines SLOs, alerts, and rollback mechanisms.

12. Q12

    Design a migration and retention plan for Amazon's payments reconciliation with emphasis on cache metadata, invalidation logs, materialized views, and read-through/write-through tradeoffs. Include tables or collections, keys, indexes, consistency guarantees, and a migration strategy

    HardDatabase DesignCaching

    How to answer:

    A strong answer starts from access patterns, then proposes online migrations, backfills, archival, data validation, and rollback. It explains transaction scope, isolation level, retention needs, and how the model handles growth to millions of merchants or hosts.

13. Q13

    Amazon's fulfillment routing has grown to millions of merchants or hosts. Propose a bottleneck reduction plan focused on hit ratio improvement, origin shielding, stampede prevention, hot-key splitting, and invalidation load while preserving limited memory per worker. Explain bottlenecks, tradeoffs, instrumentation, and rollout

    HardScalabilityCaching

    How to answer:

    A strong answer quantifies traffic first, then targets the highest-risk bottleneck with profiling, caching, batching, pooling, sharding, and load shedding. It uses staged rollout, load tests, dashboards, and fallback behavior rather than only adding more machines.

14. Q14

    Ask clarifying questions first; then answer this: Design a real-time chat service for Kindle

    HardHigh Level DesignChat service

    How to answer:

    Use conversation service, websocket gateways, message store, delivery receipts, push notifications, and idempotent message IDs; discuss ordering and offline sync. Include customer impact, operational metrics, and failure handling.

15. Q15

    Using STAR and Amazon Leadership Principles, answer this: Describe a difficult code review discussion on Retail Search

    SeniorBehavioralCode review conflict

    How to answer:

    Show you focused on maintainability and evidence, separated person from code, reached clear standards, and improved future reviews. Name the LP implicitly through actions; include metrics and ownership.

Practice these with instant AI feedback in a live mock interview → Start a Amazon Backend Engineer mock

Prepare 8-12 STAR stories mapped to Leadership Principles; expect a Bar Raiser; quantify impact

Indicative Backend Engineer pay in India: ~₹10–45 LPA (role-level range, not a Amazon-specific figure).

Other roles at Amazon

Backend Engineer interviews at other companies