TL;DR — RAG Pipeline

Design a Retrieval-Augmented Generation pipeline that ingests documents, chunks and embeds them, stores vectors for retrieval, and augments LLM prompts with relevant context to produce grounded, factual responses. The system must handle millions of documents across multiple knowledge bases. Key features: Ingest documents in multiple formats (PDF, HTML, Markdown, DOCX). Chunk documents using configurable strategies (fixed, semantic, recursive).