winkle
Back

Interview · System design

System design: backend

The candidate designs one high-volume backend service over seven connected stages, from requirements and numbers through the pipeline, failure, user limits, monitoring and scope. It suits senior and staff backend engineers. Follow-ups push for real numbers, for what happens when a dependency fails, and for the trade-off behind each choice.

Suits: Senior backend engineer, Staff software engineer, Platform engineer, Software architect

Questions
7 stages
Length
50 min
Format
Voice or typed
Levels
Senior to Staff
See this interview on a 30-minute call.
AI interviewerScored
Example

  1. Interviewer: Tell me about a time you changed someone's mind at work. How did you go about it?

  2. Candidate: Our team wanted to rebuild the billing page from scratch, and I thought it was the wrong call, so we put together a plan to…

  3. Follow-up

    Interviewer: You said "we". What did you say yourself, and what finally changed their mind?

  4. Scored on

    • Requirements and estimation20%
    • API and data design20%
    • Scale and throughput25%
    • Reliability20%

    Score, reasons and recording. Your team decides.

An example exchange. This interview's own questions open when you use it.

What it listens for, and how much each part counts.

Every answer is scored against this guide, with the reasoning written beside the score. You get the recording or the written answers too, and your team makes the call.

  • Requirements and estimation20%
    Turns the brief into clear requirements and sound back-of-envelope numbers.
  • API and data design20%
    Designs a clean, idempotent API and a data model that fits the access pattern.
  • Scale and throughput25%
    Builds a pipeline that plausibly meets the rate and explains how it scales.
  • Reliability20%
    Handles partial failure, retries and duplicates with a concrete mechanism.
  • Trade-off judgement15%
    Names what each choice costs and makes sensible cuts under pressure.

Send it alongside

All interviews

  • Coding interview: problem solving

    Engineering · Technical

    Four classic problems, solved in words, that show how a candidate reasons about correctness and cost.

    Listens for

    • Problem understanding
    • Approach and correctness
    • Edge cases
    • Complexity reasoning
    Typed

    4 of 18 questions · 35 min · All levels

  • Frontend engineering deep-dive

    Engineering · Technical

    Tests real frontend judgement: performance on weak devices, state, accessibility, rendering and debugging.

    Listens for

    • Performance
    • Architecture and state
    • Accessibility
    • Debugging
    • Trade-offs and communication
    Voice or typed

    5 of 16 questions · 25 min · Mid to Staff

  • Backend engineer

    Engineering · Technical

    Tests how a backend engineer designs data and APIs, keeps services correct and debugs them live.

    Listens for

    • Data and API design
    • Correctness under failure
    • Debugging and operations
    • Trade-off communication
    Voice or typed

    7 of 17 questions · 35 min · Mid to Senior