Design a large-scale recommendation system end to end

Design a recommendation system from scratch. When a user opens a surface, it picks a short, ordered list of items out of a very large catalog that changes all the time, and it learns from how users respond.

Walk the whole loop: framing the problem and the objective, the labels, candidate generation, ranking models and features, serving within a latency budget, cold start for new users and items, offline and online evaluation, and the feedback loops that a deployed recommender creates.