DocuMind - AI-Powered Document & Multimedia Q&A

DocumMind is a full-stack web application that allows users to upload documents (PDF) and multimedia files (Audio/Video), receive an instant AI summary, and interact with a chatbot that answers questions based on the file content.

It features a Zero-Cost architecture, leveraging local AI models (Whisper, extracted text search) to run completely free without external API keys.

---

Features

Core Functionality

• Multi-Format Support: Upload PDF documents, MP3 audio, and MP4 video files.
• Local Transcription: Uses OpenAI's Whisper model (running locally) to transcribe audio/video with high accuracy.
• Smart Contextual Chat:
  • Key-phrase matching chatbot that quotes exact sentences from your files.
  • Zero Hallucinations: Answers are strictly grounded in the document content.
• Timestamp Navigation:
  • Why read when you can listen? The chatbot provides clickable timestamps (e.g., [00:45]).
  • Jumps the built-in media player to the exact moment the topic is discussed.
• AI Summaries: Instantly auto-generates a summary preview of any uploaded file.

User Experience

• Premium UI: Glassmorphism aesthetic, dark mode, and smooth animations using Tailwind CSS.
• Drag & Drop: Intuitive file upload interface.
• Authentication: Secure Multi-User Login & Registration system (JWT-based).

---

Tech Stack

Backend

• Framework: FastAPI (Python 3.12)
• Database: MongoDB (Async Motor driver)
• Caching/Rate Limiting: Redis
• AI/ML:
  • openai-whisper (ASR/Transcription)
  • pypdf (Document Parsing)
  • argon2-cffi (Secure Password Hashing)
• Authentication: OAuth2 with JWT (JSON Web Tokens)

Frontend

• Framework: React (Vite)
• Styling: Tailwind CSS, Lucide Icons
• HTTP Client: Axios

Infrastructure

• Containerization: Docker & Docker Compose
• CI/CD: GitHub Actions

---

Getting Started

Prerequisites

• Python 3.10+
• Node.js 18+
• MongoDB (running locally on port 27017 or Docker)
• Redis (running locally on port 6379 or Docker - Required for Rate Limiting)
• FFmpeg (required for Whisper)

0. Clone the Repository

git clone https://github.com/your-username/Documind.git
cd Documind

1. Environment Setup (.env)

Create a .env file in the backend directory:

MONGO_URI="mongodb://localhost:27017"
MONGO_DB="Document"
JWT_SECRET="your_secret_key"
ACCESS_TOKEN_EXPIRE_MINUTES=1440
GROQ_API_KEY="your_groq_api_key"
REDIS_URL="redis://localhost:6379"

2. Start Services (Redis & MongoDB)

Use Docker to start the vital services:

docker-compose up -d redis mongodb

3. Backend Setup

cd backend

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run the server
uvicorn main:app --reload

Backend runs on http://localhost:8000

API Documentation (Swagger UI)

Once the backend is running, you can access the interactive API docs at:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

API Documentation Details

1. Register User

POST /auth/register

• Body:

  {
    "email": "user@example.com",
    "password": "securepassword"
  }

• Response:

  {
    "access_token": "eyJhbGciOiJIUz...",
    "token_type": "bearer"
  }

2. Chat with File

POST /chat/

• Header: Authorization: Bearer <token>
• Body:

  {
    "question": "Summarize the key points."
  }

• Response:

  {
    "answer": "Based on the document, the key points are..."
  }

3. Upload File

POST /files/upload

• Header: Authorization: Bearer <token>
• Body: multipart/form-data (Key: file)
• Response:

  {
    "filename": "report.pdf",
    "summary": "This document discusses...",
    "type": "pdf"
  }

4. Frontend Setup

cd client

# Install dependencies
npm install

# Run the development server
npm run dev

Frontend runs on http://localhost:5173

3. Docker Setup (Alternative)

Run the entire stack with a single command:

docker-compose up --build

---

Running Tests

To verify the codebase and check 95% coverage:

cd backend
pytest --cov=. --cov-report=term-missing

Manual Verifications

1. Authentication

1. Go to the app, click Sign Up.
2. Register. You will be redirected to Login.
3. Log in to access the dashboard.

2. File Q&A

1. Upload a PDF.
2. Ask a specific question about the text.
3. The bot should quote the exact sentence.

3. Video Seek

1. Upload a Video/Audio file.
2. Ask about a spoken topic.
3. Click the blue timestamp button [00:xx] to jump the player.

---