Design an Encyclopedia System Powered by User Conversation Data

Late June, 2026

README Task To one day achieve Encyclopedia Galactica, we need to start with Grokipedia--which you will build from scratch. Specifically, given a dataset of synthetic user conversations with Grok, design then implement a system that: • Ingests simulated grok.com user conversation data, • Creates encyclopedic articles on relevant topics, and • Stores them in some storage system, and • Searches stored articles so users can find them by topic For the first 30 minutes, you will design a plan for your system (no AI permitted). Interviewers will then debrief on your progress. For the next 1.5 hours, you will implement your system. You may use AI, search the web, or look up documentation — whatever helps you move fast. There are no restrictions. By the end of the interview, you should have a working encyclopedia populated with articles. At a minimum, you should be able to query the encyclopedia for articles on a given topic. However you want to implement this is up to you--be it via a CLI or a web UI. Important: The provided dataset (`queries.parquet`) is a stand-in for live production data. As you build, think about how your system would work at scale — ingesting a continuous stream of real user conversations. You're free to use web search or other external sources to enrich your articles. The conversations tell you what to write about; how you gather information and synthesize articles is up to you.