AI Engineer | RAG Pipelines · LangChain · Pinecone | Chatbot & Knowledge Base Systems

Hi, We're a small consulting firm and over the years we've accumulated around 200 PDFs (client deliverables, market research, internal SOPs, case studies). Every week someone on the team is digging through old folders trying to find an answer that's already documented somewhere. I want to fix that. What I need is a simple internal chatbot. The team types a question, the bot answers using only our documents, and it cites which PDF and which page the answer came from so we can verify it. No making things up, no answering from general knowledge — only from our docs. What I'm looking for: Set up a vector DB (Pinecone preferred, open to alternatives) Ingest all 200 PDFs with proper chunking and metadata Build the retrieval pipeline using LangChain Use OpenAI (GPT-4o or similar) for the actual answer generation Supabase for auth and storing chat history per user A simple chat UI on the web (Next.js / React is fine), internal use only FastAPI backend Answers must include source PDF name + page number Should be comfortable with Python, RAG architecture, Pinecone, LangChain, OpenAI, Supabase, and the front-end side with Next.js / React / TypeScript. I don't need anything fancy on the UI, just clean and usable. Want to wrap this in about a week. If it goes well, we'll likely have a phase 2 (Slack integration, role-based access for different teams). Skills: Python, OpenAI, Pinecone, RAG, LangChain, FastAPI, Supabase, Next.js, React, TypeScript, Vercel AI SDK