HARD
Netflix Personalization Engine
Design recommendation engine for 200M+ users, 10K+ titles, real-time viewing data.
Estimated Time: 120 minutes
#streaming#recommendations#ml#scale
Solution Overview
Hybrid recommendation: collaborative filtering + content-based. Use Kafka for real-time events, Redis for cache, DynamoDB for preferences.
Used By Companies
NetflixGoogleAmazon
Solution Overview
Hybrid recommendation: collaborative filtering + content-based. Use Kafka for real-time events, Redis for cache, DynamoDB for preferences.
Approach
Kafka real-time pipeline, ML ranking, collaborative + content-based filtering
Companies
- •Netflix
- •Amazon
Components
- •Event streaming (Kafka)
- •ML ranking
- •Cache (Redis)
- •Time series DB
- •Search engine