AI-powered document intelligence platform using RAG (Retrieval-Augmented Generation) to upload PDFs, ask questions, and generate MCQs.
DocMind AI is a full-stack application that enables users to interact with documents intelligently using modern AI techniques.
It leverages RAG (Retrieval-Augmented Generation) to:
- Retrieve relevant document chunks using embeddings
- Augment LLM prompts with context
- Generate accurate and context-aware responses
RAG (Retrieval-Augmented Generation) is a technique where:
- Documents are converted into embeddings
- Relevant chunks are retrieved based on user queries
- Retrieved context is passed to an LLM
- LLM generates answers grounded in the document
This avoids hallucination and improves accuracy.
- React.js
- Vanilla CSS
- Fetch API
- Node.js
- Express.js
- bcrypt
- multer
- jsonwebtoken
- MongoDB (Mongoose)
- OpenRouter (LLM - GPT OSS model)
- Embeddings (OpenRouter / Cohere)
- PDF Parsing
- Custom text chunking
- User uploads a PDF
- Text is extracted and split into chunks
- Each chunk → converted into embeddings
- Stored in MongoDB
- User asks a question
- Relevant chunks are retrieved
- LLM generates response using retrieved context
- JWT Authentication (Signup/Login)
- PDF Upload & Parsing
- RAG-based Question Answering
- Semantic Search using Embeddings
- AI-powered MCQ Generator
- Document Management (view/delete)
- Modern responsive UI
bash git clone https://github.com/riteshsharma2406/DocMind.git cd DocMind
bash cd backend npm install
Create .env file:
env PORT=3000 MONGO_URI=your_mongodb_uri JWT_SECRET=your_secret OPENROUTER_API_KEY=your_key
Run backend:
bash npm start
bash cd frontend npm install npm start
- Frontend: Vercel
- Backend: Render
- Database: MongoDB Atlas
- MCQ history & analytics
- Multi-document RAG querying
- Advanced vector search (FAISS / Pinecone)
- PDF preview with highlighting
- Faster embedding pipeline
Ritesh Sharma
GitHub: https://github.com/riteshsharma2406
If you like this project, give it a ⭐