Design a real-time game recommendation system
Company: Roblox
Role: Machine Learning Engineer
Category: ML System Design
Difficulty: hard
Interview Round: Technical Screen
Design a real-time game recommendation system for a platform with millions of games and tens of millions of users. Focus on the system architecture, not ML model details. Describe:
1) end-to-end request flow and latency SLOs;
2) online/offline data ingestion (user events, game metadata) and schemas;
3) streaming feature computation and a real-time feature store;
4) candidate generation and ranking service interfaces;
5) caching strategy and freshness/TTL;
6) user personalization state and session handling;
7) scalability plans (QPS estimates, partitioning, replication, storage);
8) experimentation and rollout (A/B, feature flags);
9) cold-start strategies;
10) failure modes, consistency, and backfill/lag recovery;
11) privacy, security, and abuse prevention;
12) monitoring, alerting, and cost controls.
Overview: Design a real-time game recommendation system evaluates ML product requirements, data/labeling, modeling, serving architecture, evaluation, monitoring, and trade-offs in a realistic interview setting. A strong answer states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.