# Build a Recommendation System and Conversational Interface from Scratch Design an initial recommendation system, then extend it with a chatbot that can retrieve and explain recommendations while respecting the same eligibility and ranking rules. ### Constraints & Assumptions - Start with no existing recommender infrastructure. - The chatbot must not invent unavailable items. ### Clarifying Questions to Ask - What item, feedback signals, and success metric apply? - Does conversation alter long-term preferences? ### What a Strong Answer Covers - Data, baseline retrieval/ranking, feedback, cold start, evaluation, and serving. - Grounded chatbot tools, session state, safety, and latency. ### Follow-up Questions - How would explanations be evaluated? - How would you prevent feedback loops?