BU Chatbot: Bahria University Student Smart Assistant

An intelligent, RAG-powered (Retrieval-Augmented Generation) chatbot designed to help students navigate the Bahria University Student Rulebook (Spring 2025). This project leverages MongoDB Atlas for vector search, Groq for high-speed LLM inference, and Clerk for secure user authentication.

---

🚀 Overview

The BU Chatbot transforms a static PDF rulebook into an interactive AI assistant. It uses a hybrid search approach (Regex + Vector Search) to provide accurate answers based on the official university guidelines.

Key Features

• RAG Engine: Combines semantic vector search with specific rule/page regex matching.
• Async Performance: Built with FastAPI and Motor (Async MongoDB driver) for high concurrency.
• Memory Management: Implements the Bucketing Pattern in MongoDB to store and retrieve chat history efficiently.
• Authentication: Integrated with Clerk for secure, JWT-based user session management.
• Automated Knowledge Base: Scripts to automatically download, process, and ingest university rules from Kaggle into MongoDB Atlas.

---

🛠️ Tech Stack

• Backend: FastAPI (Python)
• LLM: Groq (Llama 3.1 8B)
• Database: MongoDB Atlas (Vector Search & Chat Storage)
• Embeddings: FastEmbed (all-MiniLM-L6-v2)
• Auth: Clerk
• Dataset: Kaggle (Structured JSON)

---

📋 Prerequisites

1. Python 3.9+
2. MongoDB Atlas Account: Create a cluster and obtain your connection string.
3. Groq API Key: Get one from the Groq Console.
4. Clerk Account: Set up an application to get your API keys and JWT configuration.

---

⚙️ Installation & Setup

1. Clone the Repository

git clone https://github.com/BeUnMerreHuman/BU-Chatbot.git
cd BU-Chatbot

2. Install Dependencies

pip install -r requirements.txt

3. Configure Environment Variables

Create a .env file in the root directory and populate it with your credentials:

# Database Configuration
MONGODB_URI=your_mongodb_atlas_uri
DB_NAME=data
CHAT_BUCKET_SIZE=50
VECTOR_INDEX_NAME=vector_index
VECTOR_COLLECTION_NAME=rules

# AI Configuration
GROQ_API_KEY=your_groq_api_key

# Clerk Authentication
CLERK_PUBLISHABLE_KEY=pk_test_...
CLERK_SECRET_KEY=sk_test_...
CLERK_ISSUER=https://your-issuer-url
CLERK_JWKS_URL=https://api.clerk.com/v1/jwks

4. Initialize the Database

Run the AddData.py script. This script performs three critical tasks:

1. Downloads the BU Student Rulebook Dataset from Kaggle.
2. Initializes the MongoDB schema, including chat session and message collections.
3. Vectorizes the rules and uploads them to your Atlas Cluster.

python AddData.py

Note: You must create a Vector Search Index in MongoDB Atlas named vector_index on the rules collection to enable semantic search.

---

🏃 Running the Application

Start the FastAPI server:

python -m uvicorn main:app

The application will be available at http://localhost:8000.

---

📂 Project Structure

• main.py: The FastAPI application entry point and API endpoints.
• RAGengine.py: Logic for query contextualization, document retrieval, and LLM answer generation.
• database.py: Handles all MongoDB interactions using the Bucketing Pattern for chat history.
• AddData.py: Data ingestion pipeline (Kaggle -> Vector Store).
• static/: Contains the frontend assets (index.html, script.js, style.css).
• requirements.txt: List of Python dependencies.