Build an AI-Powered Document Q&A Web App for Internal Business Teams

We are building an internal web application that lets business teams upload documents (PDFs, Word files, CSVs) and instantly interact with them using natural language — getting AI-generated summaries, extracting key data points, and asking plain-English questions about the content. The goal is a clean, fast, production-ready tool that replaces the need to manually read through long reports, contracts, or research documents. Think of it as a private ChatGPT for your own files — fully contained, no data leaving your infrastructure. The core stack we have in mind: — Python backend handling document parsing, chunking, and the LLM pipeline (RAG-based Q&A using OpenAI or Anthropic Claude) — Node.js serving the REST API that connects the AI backend to the frontend — JavaScript/React frontend with a responsive, clean UI for document uploads, chat interface, and summary views — Supabase or PostgreSQL for document metadata and conversation history — Deployed to Railway, Render, or a similar platform We want the system to handle multi-document sessions (a user can upload several files and ask questions across all of them), track conversation history per session, and return answers with source citations (which document, which section). This is not a prototype — we expect clean code, error handling, and a deployment we can hand off to non-technical staff to use from day one. Screening questions 1. Have you built a RAG pipeline before? What vector store did you use (Pinecone, pgvector, Chroma, etc.)? 2. Share an example of a full-stack AI project you've shipped — GitHub or live link preferred. 3. Which LLM provider would you use for this project, and why? 4. What's your approach to handling large documents that exceed the context window? Skills required: JavaScript · Python · Web Development · Artificial Intelligence · Node.js