🤖 Document Assistant RAG

An intelligent document assistant that allows you to upload PDF/Text files and chat with them using Retrieval-Augmented Generation (RAG). Powered by FastAPI, React, Qdrant, and Google Gemini.

demo_upload.mp4

demo_questions.mp4

---

🚀 Key Features

• 📂 Multi-format Upload: Support for PDF and TXT files.
• 🧠 Intelligent RAG: Uses semantic search to find relevant context for your questions.
• 💬 Real-time Chat: Interactive chat interface with source citations.
• 🌓 Dark Mode: Sleek, modern design with full dark mode support.
• 🐳 Dockerized: Easy deployment using Docker and Docker Compose.

---

🏗️ Architecture

The application follows a modern RAG architecture:

graph TD
    A[User] -->|Uploads Files| B(Frontend - React)
    B -->|API Request| C(Backend - FastAPI)
    C -->|Process & Chunk| D[Document Loader]
    D -->|Embeddings| E[Google Gemini API]
    E -->|Store Vectors| F[(Qdrant Vector DB)]
    A -->|Asks Question| B
    B -->|Query| C
    C -->|Semantic Search| F
    F -->|Context| C
    C -->|RAG Chain| E
    E -->|Answer| C
    C -->|JSON| B
    B -->|Display| A

Loading

---

🛠️ Environment Variables

Create a .env file in the root directory based on .env.example:

Variable	Description	Default
GOOGLE_API_KEY	Your Google AI Studio API Key	Required
API_PORT	Port for the Backend API	8000
FRONTEND_PORT	Port for the Frontend application	5173
QDRANT_HOST	Hostname for Qdrant	qdrant
QDRANT_API_PORT	Port for Qdrant API	6333
QDRANT_GRPC_PORT	Port for Qdrant GRPC	6334

---

📦 Getting Started

1. Using Docker (Recommended) 🐳

The easiest way to run the entire stack is using Docker Compose:

# Build and start all services
docker-compose up --build

Access the applications at:

• Frontend: http://localhost:5173
• Backend API: http://localhost:8000/docs

2. Manual Setup (Development) 💻

Backend

cd backend
# Install dependencies (requires Poetry)
poetry install
# Run development server
poetry run python main.py

Frontend

cd frontend
# Install dependencies
npm install
# Start dev server
npm run dev

---

🧪 Technical Stack

• Frontend: React 19, Vite, TailwindCSS, shadcn/ui.
• Backend: FastAPI, LangChain, PyPDF, Qdrant Client.
• AI/LLM: Google Gemini (via LangChain).
• Database: Qdrant (Vector Database).
• Deployment: Docker, Docker Compose.

Note

Make sure your GOOGLE_API_KEY has access to the Gemini 1.5 models.