Description
OSI is an intelligent virtual assistant designed for IT support at the French Ministry of Economy and Finance. The project consists of three main components: a RAG (Retrieval-Augmented Generation) based backend API, an interactive chatbot frontend, and an administration dashboard aimed at continuously improving the chatbot and accessing usage statistics.
Project Components
Backend API
- RAG system coupled with LLM for contextual responses
- FAISS vector database for semantic search
- REST API with WebSocket support
- Conversation and history management
- Support for multimedia content (videos and images)

Frontend Chatbot
- Interactive and responsive chat interface
- Dark/light mode
- Support for messages, images, and videos
- Interactive navigation buttons
- Response rating system

Admin Dashboard
- Secure administration interface
- RAG knowledge base management
- Usage statistics visualization
- System configuration
- User message management

Message Management

RAG Knowledge Base

System Configuration

Usage Statistics


Main Technologies
- Backend: Python, FastAPI, FAISS, MongoDB
- Frontend: Next.js, Tailwind CSS, Framer Motion
- AI: LLM (Mistral), Embeddings (Solon)
- Infrastructure: Docker, WebSocket
- Database: MongoDB, FAISS
Key Features
- Semantic search in knowledge base
- Contextual response generation
- Modern and responsive user interface
- Secure authentication system
- Usage monitoring and statistics
- Complete knowledge base management