ML Search / RAG System Design

Problem Overview

Design a Retrieval-Augmented Generation (RAG) system, typically framed as an enterprise-grade system that allows employees to ask questions over internal documents (similar to Glean). The candidate must walk through all major components: document ingestion and chunking, text embedding, vector storage and retrieval, reranking, and the LLM-based generator. The interviewer expects…

• 6 candidate-reported follow-ups — the exact probes interviewers asked, with the trigger for each
• The rest of the problem statement — full requirements, constraints, and edge cases
• Approach and trade-offs — what passing candidates did, and the mistakes that sink people