Backend engineer interviews lean on three pillars: API design, database internals, and distributed systems. The depth expected scales sharply with level. A mid-level role will ask you to design a paginated REST API; a senior role will ask you to handle hot-key cache invalidation across multi-region replicas. Knowing the canonical question patterns means you spend the round driving the discussion instead of being driven.
API Design
Pagination — Cursor vs Offset
Be ready to explain why cursor-based pagination is preferred over offset for large datasets (stable ordering under inserts, O(1) lookup with an indexed cursor). Expect to defend the trade-off — random-page access is harder with cursors.
REST vs gRPC vs GraphQL
There is no universal answer. The right answer depends on consumer (browser vs internal service), schema evolution needs (gRPC's protobuf is best-in-class), and latency tolerance. The candidates who say 'GraphQL is always better' lose points.
Idempotency Keys
When asked how to handle retries safely, mention idempotency keys, the difference between idempotent operations and at-most-once delivery, and how Stripe's payment API solves the duplicate-charge problem.
Databases
Indexes — What and When
Be ready to explain why a B-tree index helps point queries and range scans but not LIKE '%foo%' queries (no left-anchor). Be ready to talk about covering indexes and the cost of write amplification when you over-index.
Sharding Strategies
Hash-based, range-based, geo-based — know the trade-offs. Hash is balanced but breaks range queries; range is range-query-friendly but creates hotspots. Senior loops will probe the failure mode of each.
Read Replicas and Consistency
Expect 'a user updates their profile and immediately fetches it — they see the old value. Why?' (Replication lag.) Be ready to talk about read-your-own-writes, monotonic reads, and how to mitigate (sticky sessions, read-from-primary for the affected user).
Distributed Systems (Senior+)
CAP, PACELC, and Trade-offs
Do not recite CAP — apply it. "Network partition between regions, do you favour consistency or availability?" depends on the use case. Payments → consistency. Like-button → availability. Senior loops want concrete reasoning, not theory recitation.
InterviewLift's backend track has a system-design rubric calibrated against the exact level expectations at Google, Meta, Stripe, and Amazon — so you know whether your design answer is hitting L4, L5, or L6 before the loop tells you.
Reviewed by the InterviewLift customer-success team, who prepare candidates for real interviews every week.
Try Jarvis free: 15 questions
No card needed. Use it in your next practice round or interview.