System design interview question
Design a Distributed Recommendation Engine and Real-Time Personalization System for a Global Streaming Platform (Like Netflix's Content Recommendation at Scale)
Design a distributed recommendation system serving personalized content suggestions to 200M active users across 6 continents, handling 50K recommendation requests/second with <200ms latency. The system must ingest real-time user behavior (views, clicks, ratings) from billions of daily events, compute personalized recommendations using collaborative filtering and content-based models, and serve diverse ranking strategies (trending, personalized, cold-start) while maintaining consistency across regional data centers. Key constraints include: handling 5 petabytes of historical user interaction data, supporting online model inference with sub-100ms compute, maintaining feature freshness (<5 min staleness), surviving single-region failures without service degradation, and supporting A/B testing across recommendation algorithms without cache coherency issues.
Practice this question step by step with an AI interviewer that reviews your requirements, your architecture and your trade-offs, and scores the design.