Backend Developer Interview Questions: What Gets Asked

Backend job descriptions list a dozen technologies, and it is hard to tell which ones the interview will actually test. This section maps the question areas that recur across backend loops. The stack changes from company to company, but the underlying topics stay stable.

• API design. REST versus RPC, versioning, pagination, error responses, authentication, and idempotency keys.
• Databases and data modeling. Choosing relational or document storage, indexes, joins, transactions, isolation levels, N+1 queries, and running schema migrations safely.
• Caching and performance. Cache invalidation, expiry, cache stampedes, and how you would profile a slow endpoint.
• Concurrency and reliability. Race conditions, locks, queues, timeouts, and retries with backoff.
• System design. Prompts such as a URL shortener, a rate limiter, or a notification service.
• Coding. Data structures and a practical problem, usually in the language of your choice.

Language and framework rounds go deeper on specifics; the question banks hub collects sets for Go, Java, Spring Boot, microservices and more.

These questions appear in some form in most backend loops. Practice answering each one aloud in under two minutes:

• What happens when you add an index to a column, and when does an index hurt?
• What is the difference between read committed and serializable isolation, and when would you pay for the stricter level?
• How would you find and fix an N+1 query in an ORM?
• How do you make a payment endpoint safe for a client to retry?
• How do you paginate a large result set that changes while the user scrolls?
• When would you choose a document store over a relational database?
• How would you roll out a column rename without downtime?

A strong answer follows the same shape each time: state your default choice, name the cost of that choice, and describe the situation where your answer would flip. Interviewers follow up on the flip case, so have a real example ready.

For system design, start by clarifying requirements: who calls the service, what must never be lost, and which reads and writes dominate. Then sketch the simplest design that works, and only add caches, queues or sharding when you can say which requirement forces them.

Debugging questions are where real experience shows. A backend engineer interviewing for a mid-level role at a payments company was asked why an endpoint's latency doubled after a release. Instead of guessing at causes, they described their steps: check what changed in the deploy, compare traces for the slow path, read the query plan, and confirm the fix against the same latency metric. They then tied it to a slow-query incident from their previous job, and the rest of the conversation was about that incident rather than about tools.

Prepare two production stories before any backend interview: a bug you tracked down and a design decision you would now make differently.

Reading answers is not the same as saying them. Run the question list above aloud against a timer, or let the mock interview tool ask the questions and follow up on vague answers.

For the live video interview itself, SubcueAI transcribes the interviewer and drafts structured answer points in a local overlay, either in the desktop app on macOS or Windows (Zoom, Google Meet, Microsoft Teams) or in the browser extension's Side Panel for calls in a Chrome or Edge tab. No bot joins the call. It is most useful in the conversational rounds, such as design discussion and experience questions. Coding rounds on a shared screen, proctored assessments, recorded one-way interviews and company-managed devices are outside what it can safely cover, and anything you share on screen is visible. Setup is on the tutorial page.