Your question is Distributed Embedding and Vector Search. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
Design a distributed embedding generation and vector search service capable of handling millions of dynamic documents.
Clarify the required document update and query patterns, then design the ingestion, embedding, indexing, and search architecture. Cover online versus batch processing, consistency and freshness, scaling, cost, failure handling, evaluation, monitoring, and training-serving skew. State reasonable assumptions where requirements are unspecified.
