Personalized Product Feed for an Online Store

Reported in interviews at Atlassian Shopify

Problem statement

How would you build a personalized product feed for a large online store? Traffic and the catalog are large, stock and prices change often, and the feed must load quickly. The current feed shows category bestsellers, but the goal is more orders with available, varied and fresh products. It must also serve new users and products.

Explain training examples, the ranking pipeline and cold-start retrieval.

Show hint

Work out the request rate at peak and what it would cost to score the whole catalog; that decides how many stages you need and where availability must be enforced.

Relevant roles: ML Engineer, Applied Scientist and AI Engineer.

Related free problems