#MSCS Academic RAG Chatbot

An intelligent academic chatbot for Rowan University MSCS students that answers questions based on uploaded curriculum documents with exact page citations.Built using a Retrieval-Augmented Generation (RAG) pipeline.

##What It Does

• Students ask questions about their MSCS curriculum
• Chatbot finds the most relevant sections from uploaded PDFs
• Answers are grounded only in the curriculum — no hallucination
• Every answer shows exactly which page the information came from
• Falls back to general CS knowledge when topic isn't in the curriculum
• Handles greetings and casual conversation naturally

How It Works (RAG Pipeline)

Student Question ↓ Convert question to vectors (Gemini Embeddings) ↓ Search FAISS vector store for relevant chunks ↓ Retrieve top 3 most relevant curriculum sections ↓ Send question + retrieved chunks to Groq (Llama 3) ↓ Answer grounded in YOUR curriculum + page citations

Tech Stack

Backend:FastAPI + Python Vector Search:FAISS Embeddings:Google Gemini (text-embedding) LLM:Groq(Llama 3.3 70B) PDF Processing:PyPDF2+tiktoken Frontend:HTML+CSS+Vanilla JavaScript

Setup & Run

1. Clone the repository

git clone https://github.com/chandupragada/rag-based-chatbot.git cd rag-based-chatbot/backend

2. Create virtual environment

python -m venv venv venv\Scripts\activate # Windows source venv/bin/activate # Mac/Linux

3. Install dependencies

pip install -r requirements.txt

4. Set up API keys

Create a .env file inside backend/ folder: GEMINI_API_KEY=your_gemini_key_here GROQ_API_KEY=your_groq_key_here

Get Gemini key free at: https://aistudio.google.com Get Groq key free at: https://console.groq.com

5. Add your curriculum PDFs

Put your PDF files inside backend/data/ folder

6. Ingest PDFs into knowledge base

python ingest.py

7. Start the backend server

python main.py

8. Open the chatbot

Open frontend/index.html in your browser Make sure backend server is running first!

Cost

This project runs on free tier APIs only no credit card needed!