Uber/Airbnb-Style Vector Search Interview (EXPLAINED)
Company-based system design: sharded ANN, metadata pre-filtering, query routing, reranking, and latency SLOs at marketplace scale.
TL;DR — Quick Answer
Shard indexes by region or marketplace vertical, embed listings with text+image multimodal vectors, pre-filter on geo/metadata before ANN, scatter-gather top-k, rerank with business signals (conversion, availability), and cache hot queries. Target sub-100ms p99 with async index updates acceptable for non-critical freshness.
The Interview Question
Design vector search for a marketplace with millions of listings, geo filters, and personalized ranking — Uber/Airbnb-style interview.
Deep Explanation
Interview prompt
Sign in to unlock full answer
Get deep explanations, PDF export & all Vector Databases questions
- 25 more sections of deep explanation
- Real-world examples
- Common mistakes
- Interviewer expectations
- Follow-up questions
UberAirbnbMarketplaceSystem DesignUberAirbnb