Exercise · Search & Discovery

Search Autocomplete

Whiteboard exercise. Try the problem cold, then reveal the rubric to self-score.

Out of 10 points45 min whiteboardReference solution →
01

Prompt

A typeahead system that returns ranked suggestions on every keystroke — at sub-100ms latency, for billions of daily searches. The challenge isn't finding matches. It's pre-organizing data so matches are already waiting.

Time budget: 45 min whiteboard. Draw architecture, estimate numbers, discuss tradeoffs.

02

Hints (progressive — click to reveal)

Hint 1

Start with requirements: functional vs non-functional. Clarify the scale (users, QPS, storage).

Hint 2

Think about the data model first. What entities exist? What are the access patterns?

Hint 3

Identify the hardest sub-problem and deep-dive into it. Show you can go beyond boxes and arrows.

03

Rubric — 10 points

  • +2 Back-of-envelope estimation with concrete numbers
  • +2 Clear API design with key endpoints
  • +2 Sensible data model and storage choices
  • +2 Addresses scalability (sharding, caching, CDN)
  • +2 Discusses failure modes and mitigations

Self-score: tally the points you would have mentioned unprompted. 7+ is interview-ready on this problem.

04

Red flags (things that tank the interview)

  • Scan full text of all suggestions per keystroke
  • Hit your authoritative DB on every keystroke
  • Use generic search index (ElasticSearch) for autocomplete