RAG System Developer — Enterprise Knowledge Base

Situation: 8 years of consulting reports. Thousands of PDFs. Analysts spend 40 minutes searching before they start writing anything new — often duplicating work that already exists in a document from 2021 that nobody can find. Goal: a RAG-powered internal search tool that our analysts can query in plain English and get relevant excerpts back within seconds. Not a chatbot. Not a full product. Just something that ingests our documents, retrieves well, and has a simple enough interface that a non-technical analyst can use it on day one. Deliverables: – Ingestion pipeline (PDFs, .docx, email exports) – Vector store (Pinecone, Weaviate, or whatever you think is right) – Simple chat interface – Evaluation setup so we can measure retrieval quality – Full documentation — this will be maintained internally after handover