Question 1

Tell me about a challenging bug you fixed.

Accepted Answer

Situation: A nightly data job started failing intermittently in staging. Task: I owned the pipeline and needed it stable before launch. Action: I added structured logging, reproduced the race condition locally, and found two workers writing the same file; I introduced a lock and a retry with backoff. Result: Failures dropped to zero over two weeks and I documented the fix so the team could reuse the pattern.

Question 2

Describe a time you disagreed with a teammate on a technical decision.

Accepted Answer

Situation: A teammate wanted to add a new dependency for date handling. Task: I felt our existing utilities covered it. Action: Instead of vetoing, I prototyped both, compared bundle size and maintenance cost, and shared the numbers. Result: We kept the lightweight option, and the teammate appreciated that I argued with evidence rather than opinion.

Question 3

Tell me about a project you are proud of.

Accepted Answer

Situation: My capstone team built a campus ride-share app. Task: I led the backend. Action: I designed the API, set up CI, and paired with teammates who were newer to Git. Result: We shipped on time, onboarded 200 test users, and I learned how much mentoring accelerates a team.

Question 4

How would you find a duplicate number in an array?

Accepted Answer

I would clarify constraints first: size, value range, and whether I can mutate the input. A hash set gives O(n) time and O(n) space. If memory is tight and values are bounded, I can sort or use cycle detection (Floyd's) for O(1) space. I would state the trade-off and pick based on the interviewer's constraints.

Question 5

Explain the difference between a process and a thread.

Accepted Answer

A process has its own isolated memory space; threads share the memory of their parent process. Threads are cheaper to create and communicate, but shared state introduces race conditions you must guard with locks. I would mention a real example, like using threads for concurrent I/O but isolating untrusted work in a separate process.

Question 6

What happens when you type a URL into a browser and hit enter?

Accepted Answer

DNS resolves the domain to an IP, a TCP (and TLS) connection is established, the browser sends an HTTP request, the server responds with HTML, and the browser parses it, requests assets, builds the DOM, and renders. I can go deeper on any layer the interviewer is interested in.

Question 7

How do you decide between SQL and NoSQL for a project?

Accepted Answer

I start from the access patterns and consistency needs. Relational data with complex joins and strong transactional guarantees points to SQL. High write throughput, flexible schemas, or simple key lookups at scale point to NoSQL. For most early-stage products I default to a relational database until a clear access pattern justifies otherwise.

Question 8

What is the time complexity of common operations on a hash map?

Accepted Answer

Average case insert, lookup, and delete are O(1); worst case is O(n) when many keys collide into one bucket. Good hash functions and resizing keep it near constant. I would mention that ordering is not guaranteed unless I use an ordered map variant.

Question 9

How would you design a URL shortener?

Accepted Answer

I would scope it: generate a short key, redirect on lookup, and handle scale. I would use a base62 encoding of an incrementing ID or a hash, store the mapping in a key-value store with a cache in front for hot links, and add analytics asynchronously. I would discuss collision handling and read-heavy scaling.

Question 10

How would you design a rate limiter?

Accepted Answer

I would clarify limits (per user, per IP) and tolerance for bursts. A token bucket allows controlled bursts and is simple to reason about; a sliding window is more precise. I would store counters in a fast in-memory store like Redis with atomic increments, and fail open or closed depending on the endpoint's risk.

Question 11

How do you keep a service available when a dependency is slow?

Accepted Answer

I would add timeouts so slow calls don't pile up, a circuit breaker to stop hammering an unhealthy dependency, and a fallback (cached data or a degraded response). I would also bulkhead resources so one slow dependency can't exhaust all threads.

Software Engineer Interview Questions & Sample Answers (2026)

Behavioral Questions