🧠 YouTube RAG Chatbot Chrome Extension

A Chrome Extension that allows users to ask questions about any YouTube video and get AI-powered answers — using Retrieval-Augmented Generation (RAG), YouTube transcripts, HuggingFace embeddings, FAISS, and Gemini Pro (Google Generative AI) via a FastAPI backend.

---

📌 Features

• ✅ Chrome extension to interact with any YouTube video
• 🎙️ Fetches video transcript via YouTube Transcript API
• ✂️ Splits and embeds content using HuggingFace Transformers
• 💾 Stores vector data using FAISS
• 🧠 Sends context-aware queries to Gemini Pro
• ⚡ Uses a FastAPI backend for serving responses
• 📥 Provides real-time answers in the extension popup

---

🛠 Tech Stack

Component	Tool/Library
Language Model	Gemini Pro via langchain_google_genai
Embeddings	sentence-transformers/all-MiniLM-L6-v2
Vector DB	FAISS (langchain_community.vectorstores)
Transcripts	youtube_transcript_api
Backend	FastAPI
Frontend	HTML, JS, CSS (Chrome Extension)

---

🚀 Getting Started

🔧 1. Backend Setup (FastAPI + LangChain + Gemini)

# Clone the repo
git clone https://github.com/your-username/youtube-rag-chatbot.git
cd youtube-rag-chatbot/backend

# Setup Python 3.10 virtual environment
python3.10 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt
GOOGLE_API_KEY=your_google_gemini_api_key
uvicorn main:app --reload

🌐 2. Chrome Extension Setup

Go to chrome://extensions/ in Chrome

Enable Developer Mode

Click Load Unpacked

Select the /extension folder from the project directory

Open a YouTube video and launch the extension

💡 How It Works

Extract video ID from YouTube URL

Use youtube_transcript_api to fetch the transcript

Split the transcript into chunks using RecursiveCharacterTextSplitter

Embed each chunk with HuggingFaceEmbeddings

Store in FAISS vector index

On user question → retrieve top relevant chunks

Format context → query Gemini with langchain_google_genai

Return answer → display in popup

🤝 Credits

LangChain

HuggingFace

Gemini Pro

FAISS by Facebook

YouTube Transcript API

📄 License

This project is licensed under the MIT License.