Product-based Java interviews

Also called: product companies, consumer tech, big tech

Data structures and algorithms first, and often decisively. The interview is calibrated to a hiring bar rather than to a project vacancy, so the questions are unfamiliar by design and the interviewer is watching how you approach a problem you have not seen. Java-specific knowledge matters, but it is tested as depth on a few things rather than breadth across many. Two things carry unusual weight here relative to the other types: how you communicate while solving, and behavioural rounds.

AmazonGoogleMicrosoftFlipkartSwiggyZomatoRazorpayPhonePeAtlassianUber

The rounds

  1. 1. Online assessment

    90-120 min

    Two to three DSA problems, medium to hard, on a judged platform. Some companies add a debugging or workstyle section.

    Clears it: Fully passing test cases, including edge cases, within the timer.

    Ends it: A brute-force solution that times out on the large inputs.

  2. 2. DSA round 1

    45-60 min

    Live coding with an interviewer. Arrays, strings, hashing, two pointers, trees. You are expected to talk through the approach and complexity before writing.

    Clears it: Stating the approach and its complexity, then coding it cleanly.

    Ends it: Silent thinking, or jumping to code and discovering the approach is wrong.

  3. 3. DSA round 2

    45-60 min

    Harder — graphs, dynamic programming, heaps, or a design-flavoured data structure problem such as an LRU cache.

    Clears it: Recognising the pattern quickly and handling follow-up constraints.

  4. 4. Machine coding or low-level design

    60-120 min

    Build a small working system — a parking lot, a rate limiter, a splitwise clone — with clean classes, no framework, and running code. Common at Indian product companies specifically.

    Clears it: Working code with sensible abstractions and no god class.

    Ends it: Unrunnable code, or over-engineering into patterns nothing needed.

  5. 5. System design

    45-60 min

    Usually from three or four years upward. Scale estimation, data modelling, storage and caching choices, and explicit trade-offs.

    Clears it: Driving the conversation and naming what you are trading away.

  6. 6. Behavioural / bar raiser

    45-60 min

    Structured questions about past situations, often mapped to published company values. Amazon's Leadership Principles are the clearest example. Answers are expected in STAR form with your specific contribution and real numbers.

    Ends it: Answering as 'we' throughout, so nothing you personally did is visible.

What it weights

Relative emphasis out of five — where preparation time actually pays back here.

  • DSA

    The primary filter at every level. Insufficient DSA is not offset by strong Java.

  • Problem-solving communication

    Graded explicitly. A correct silent solve can still fail the round.

  • Low-level design / machine coding

    Particularly at Indian product companies from two years upward.

  • System design

    From roughly three years, and dominant at six-plus.

  • Behavioural

    A real rejection reason here, unlike at most service firms.

  • Java depth

    Fewer topics, much deeper — collections internals, concurrency, the JVM.

By experience

The same company type asks a different interview at two years and at six. Questions that link have a full write-up.

1-2 years

DSA is nearly the whole bar

Two or three algorithm rounds and one Java-fundamentals conversation. Java depth expectations are modest; the algorithm bar is not lowered much for juniors.

Where to spend your time

Questions at this band— 3 answered in depth

3-4 years

DSA plus low-level design and real Java internals

Machine coding or LLD becomes a distinct round, an easier system design discussion appears, and Java questions go to internals — concurrency, collections implementation, memory.

Where to spend your time

Questions at this band— 1 answered in depth

  • Design an LRU cache with O(1) get and put.
  • How does ConcurrentHashMap achieve thread safety?
  • Is HashMap thread-safe on modern Java? The infinite loop was fixed.
  • What does volatile guarantee, and what does it not?
  • When is parallelStream() a mistake?
  • Design a rate limiter, then make it distributed.
  • When would you use LinkedList instead of ArrayList?

5-6 years

System design carries the loop

DSA remains a filter but stops being the differentiator. Design depth, trade-off reasoning, production judgement and behavioural signal decide the outcome.

Where to spend your time

Questions at this band

  • Design a URL shortener for 100 million writes a day.
  • How do you size and shut down a thread pool correctly?
  • How would you diagnose rising p99 latency with normal CPU?
  • What are virtual threads, and when do they not help?
  • How do you choose between strong and eventual consistency here?
  • Tell me about a time you disagreed with a technical decision.
  • How do you find and fix a memory leak in a live service?

Why candidates get rejected here

  • A working solution that is asymptotically worse than required
  • Solving silently, so the interviewer cannot assess your reasoning
  • Machine-coding submissions that do not run
  • Behavioural answers with no personal ownership or no measurable outcome
  • Not asking clarifying questions before designing

Compare with

Read this as a guide, not a prediction. These are patterns across many interviews, not a guarantee about any one company. Process varies by company, by team, by hiring manager and by year, and a single company can run all three shapes for different roles. Use this to decide where to spend preparation time, not to predict a specific loop.