Whenever you're ready, walk me through how you'd approach this.
How would you architect a RAG (Retrieval-Augmented Generation) pipeline for an enterprise customer using Vector Search and MosaicML?
Discuss the end-to-end design, including document ingestion and chunking, embedding generation, indexing, retrieval, prompt construction, MosaicML model selection or fine-tuning, serving, security, evaluation, and monitoring. Clarify missing scale, latency, freshness, tenancy, and compliance requirements before making capacity or model recommendations.