This question evaluates competency in designing large-scale short-video recommendation systems, including machine learning model selection, candidate generation and ranking, real-time personalization, feedback signal design, evaluation metrics, latency and scalability constraints, cold-start handling, exploration–exploitation trade-offs, and safety/abuse controls. It is commonly asked in the ML system design domain to assess system-level machine learning engineering and product-aware architectural thinking, and it combines conceptual understanding with practical application by requiring both high-level trade-off reasoning and concrete serving and evaluation considerations.
Design an end-to-end recommendation system for a short-video feed product. The system serves a large user base and must choose and rank videos for each user session with low latency. Discuss the product goal, training data and feedback signals, candidate generation, ranking, real-time personalization, cold start for new users and new creators, exploration vs. exploitation, online serving architecture, safety and abuse controls, and how you would evaluate the system offline and online.