Design a Typeahead Suggestion System
Company: Pinterest
Role: Software Engineer
Category: System Design
Difficulty: easy
Interview Round: Onsite
Design a typeahead system that suggests completions as a user types. Begin by clarifying the suggestion corpus, matching and ranking behavior, freshness needs, and expected query load; the interview source specifies the system-design topic without fixed constraints.
Explain both the serving path and how the suggestion index is built and updated.
### What a Strong Answer Covers
- A clear distinction between finding eligible completions and ranking them.
- An index suited to prefix lookup and a bounded response.
- Caching, distribution, hot prefixes, and index updates.
- Consistent text handling and client behavior when requests complete out of order.
- Evaluation of usefulness, latency, and stale or inappropriate suggestions.
### Follow-up Questions
- What changes when a popular suggestion must disappear immediately rather than wait for the next index rebuild?
- How would you prevent a response for an older prefix from replacing suggestions for the user's current input?
Overview: Design typeahead with prefix indexing, ranked completions, caching, controlled index updates, text consistency, and protection against stale client responses.