BDL ChatBot (Internship Project)

A completely offline, secure, multi-user Retrieval-Augmented Generation (RAG) chatbot designed for the Bharat Dynamics Limited (BDL) office environment. It allows users to query uploaded documents locally without sending sensitive data to external servers.

Features

• Multi-User Local Server: Supports up to 5 concurrent users and 1 admin session simultaneously.
• Role-Based Access:
  • Users can chat with the AI and ask questions about the indexed documents.
  • Admins have access to a dashboard to upload documents, manage indexed files, and view system-wide chat history.
• Privacy-First (100% Offline): Uses Ollama to run Large Language Models (LLMs) entirely locally on your hardware.
• Fast & Accurate Document Retrieval: Built with PyMuPDF for high-speed PDF extraction and LanceDB for hybrid (semantic + keyword) vector search.
• Web Interface: Accessible from any browser on the local network. No client installation required.

Prerequisites

1. Python 3.10+
2. Ollama: Must be installed and running.

   # Pull the required models before starting
   ollama pull gemma4:e2b
   ollama pull nomic-embed-text

Setup & Installation

1. Clone or download the project.
2. Install the required Python dependencies:

   pip install -r requirements.txt

3. (Optional) Configure the environment variables in the .env file:
   • MODEL_NAME: The Ollama model to use (default: gemma4:e2b).
   • SERVER_PORT: Port for the web server (default: 8550).
   • MAX_USERS, MAX_ADMINS: Adjust concurrent connection limits.

How to Run

Option 1: Using the batch script (Windows) Simply double-click the start_server.bat file. It will check if Ollama is running and start the server automatically.

Option 2: Using the command line

# Start the web server (accessible on local network)
python scripts/frontend.py

# Alternatively, run as a local desktop app (single user)
python scripts/frontend.py --desktop

Once running, the terminal will display the local and network URLs (e.g., http://192.168.x.x:8550). Share the network URL with colleagues so they can connect via their web browsers.

Usage

• Default Admin Password: Admin321! (can be changed in scripts/frontend.py)
• Users can log in, select specific documents to focus on, and ask questions.
• Admins can upload PDFs, TXTs, or MD files, which the system will automatically chunk, embed, and index for querying.