Design Problems/Recommendation System

Recommendation System

Hard
MLCollaborative FilteringFeature Store

High-Level Architecture

Rendering diagram...

Design a recommendation engine (like Netflix recommendations, Amazon product suggestions, or YouTube video recommendations) that provides personalized content suggestions to millions of users in real-time. The system must balance relevance, diversity, freshness, and computational cost.

Clarifying Questions

  • What are we recommending? (Assume: products/items to users)
  • How many users and items? (500M users, 100M items)
  • How fresh do recommendations need to be? (Reflect user actions within 5 minutes)
  • Cold-start handling? (New users with no history)
  • Latency requirement? (< 200ms to serve recommendations)