🧠 RAG PDF Chatbot – LangChain + FAISS + Groq

This project is an interactive PDF chatbot that lets you upload a document and ask natural-language questions based on its content. It uses semantic search + large language models (LLMs) to return contextual answers.

---

🔁 Evolution

✅ Phase 1 – Local Ollama Setup

Initially, the chatbot used Ollama with the phi3:mini or llama3.2 model for local inference. It worked well but had heavy memory requirements and limited portability.

🌐 Phase 2 – Migrated to Groq + LLaMA 3 (Cloud)

To improve speed, reduce RAM usage, and make it deployment-ready, the app was updated to use Groq’s blazing-fast LLMs (like llama3-70b-8192) via their free API.

---

🚀 Features

• 📄 Upload any PDF document
• 🔎 Perform semantic search with FAISS
• 💬 Ask natural questions — get smart, contextual answers
• ⚡ Powered by Groq-hosted LLaMA 3 model
• 🧠 Uses Hugging Face sentence embeddings
• 🎈 Built with Streamlit for a fast, responsive UI

---

🛠️ Tech Stack

Layer	Tool/Library
Frontend	Streamlit
RAG Engine	LangChain + FAISS
Embeddings	sentence-transformers (MiniLM)
LLM Backend	Groq API (llama3-70b-8192)
PDF Processing	PyPDF2

---

🔐 Setup

1. Clone this repo

git clone https://github.com/ShallenCrissle/rag_document_chatbot.git

cd rag_document_chatbot

2.Create and Activate a Virtual Environment

python -m venv venv venv\Scripts\activate # Windows

4. Install Dependencies

pip install -r requirements.txt

4. Create .env for groq api key

5.Run the app

streamlit run app.py