📈 Document QnA AI (MERN + OpenAI + MongoDB Atlas + AWS S3)

An AI-powered web app where users can upload documents (PDF/DOCX/TXT), and then ask questions about the content. The app uses OpenAI embeddings + Chat models to answer questions based on the uploaded document, with support for file storage on AWS S3, vector search in MongoDB Atlas, and a clean React (Vite) frontend.

---

🚀 Features

• 🔐 User Authentication (Register/Login with JWT)
• 📄 Upload Documents (PDF, DOCX, TXT)
• ☁️ AWS S3 Integration — secure file storage
• 🧩 Text Extraction + Chunking
• 🧠 Embeddings with OpenAI
• 🔎 Vector Search (MongoDB Atlas) with fallback to cosine similarity
• 💬 Chat with Your Documents (QnA with citations)
• 📜 History Page — view past questions and answers
• 📂 My Docs Page — view all uploaded documents

---

🛠️ Tech Stack

Frontend: React (Vite), TailwindCSS Backend: Node.js, Express.js, MongoDB Atlas AI/ML: OpenAI API (Embeddings + GPT-4o for QnA) Storage: AWS S3 Auth: JWT (JSON Web Token)

---

📂 Project Structure

doc-qna-ai/
  ├── client/       # React (Vite) frontend
  ├── server/       # Express backend
  ├── .gitignore
  ├── README.md

---

⚙️ Setup Instructions

1. Clone the repository

git clone https://github.com/AditSaxena/doc-qna-ai.git
cd doc-qna-ai

2. Install dependencies

Backend

cd server
npm install

Frontend

cd ../client
npm install

3. Setup environment variables

Create a .env file in server/ with:

PORT=5001
MONGO_URI=your_mongodb_uri
MONGO_DB=docqna
OPENAI_API_KEY=your_openai_api_key
OPENAI_CHAT_MODEL=gpt-4o
USE_MONGO_VECTOR_SEARCH=false

AWS_ACCESS_KEY_ID=your_aws_key
AWS_SECRET_ACCESS_KEY=your_aws_secret
AWS_REGION=ap-southeast-2
AWS_S3_BUCKET=doc-qna-bucket-123

JWT_SECRET=your_jwt_secret
JWT_EXPIRES_IN=7d

4. Run the app

Backend

cd server
npx nodemon index.js

Frontend

cd client
npm run dev

---

🌟 Usage

1. Register/Login to your account.
2. Upload a document (PDF/DOCX/TXT).
3. Ask questions in the chat page.
4. View past QnA in the History Page.
5. Manage uploaded docs in My Docs Page.

---

📸 Screenshots (to be added)

• 🔑 Login/Register
• 📄 Upload Page
• 💬 Chat Page with answers & sources
• 📜 History Page

---

📦 Deployment

• Backend → Render / Railway / Heroku
• Frontend → Vercel / Netlify
• Database → MongoDB Atlas
• File Storage → AWS S3

---

📌 Resume Value

This project demonstrates:

• Full-stack MERN development
• Cloud integrations (MongoDB Atlas, AWS S3)
• Secure authentication (JWT)
• Real AI application (Document QnA with OpenAI)

---

📄 License

MIT License © 2025 Adit Saxena