🎥 YouTube RAG Chatbot

An end-to-end AI-powered Retrieval-Augmented Generation (RAG) chatbot that allows users to ask questions from YouTube video transcripts in real time.

The application extracts YouTube transcripts, processes them into semantic chunks, generates embeddings using Google Gemini, stores them in a FAISS vector database, and retrieves relevant context to generate accurate answers using an LLM.

🚀 Live Demo

🔗 Open Live App

📌 Features

• 🎥 YouTube transcript extraction
• 🧠 Retrieval-Augmented Generation pipeline
• ✂️ Semantic text chunking
• 🔎 FAISS vector database for similarity search
• 🤖 Gemini-powered embeddings and answer generation
• 💬 Interactive Streamlit user interface
• 🐳 Dockerized application
• ☸️ Kubernetes deployment configuration
• ⚙️ GitHub Actions CI workflow
• 🌐 Live deployment on Streamlit Cloud

🏗️ Project Architecture

User enters YouTube URL
        |
        v
Fetch video metadata and transcript
        |
        v
Split transcript into semantic chunks
        |
        v
Generate embeddings using Google Gemini
        |
        v
Store embeddings in FAISS vector index
        |
        v
Retrieve relevant transcript chunks
        |
        v
Generate context-aware answer using Gemini LLM

🛠️ Tech Stack

Python | LangChain | Streamlit | Google Gemini | FAISS | YouTube Transcript API | Docker | Kubernetes | GitHub Actions

📊 Project Highlights

• Built a 4-stage RAG pipeline: metadata fetch, transcript loading, FAISS indexing, and Gemini-based answer generation.
• Integrated Google Gemini embeddings and Gemini LLM for context-aware Q&A.
• Used FAISS vector search for semantic retrieval from YouTube transcripts.
• Developed an interactive Streamlit UI for real-time user queries.
• Achieved ~85-90% relevant response quality on manually tested transcript Q&A queries.
• Reduced manual video review time by enabling direct querying of long-form video content.
• Dockerized the application for consistent development and deployment environments.
• Added Kubernetes deployment and service configuration for cloud-native readiness.
• Implemented CI/CD automation for streamlined build, testing, and deployment workflows.
• Deployed the project as a live public application on Streamlit Cloud.

📂 Project Structure

YouTube-RAG-Chatbot/
│
├── app.py
├── main.py
├── requirements.txt
├── Dockerfile
├── README.md
├── .dockerignore
├── .gitignore
│
├── k8s/
│   ├── deployment.yaml
│   └── service.yaml
│
└── .github/
    └── workflows/
        └── ci.yml

⚙️ Installation

Clone the repository:

git clone https://github.com/DevaChaudhari/YouTube-RAG-Chatbot.git
cd YouTube-RAG-Chatbot

Create and activate a virtual environment:

python -m venv venv

# Windows
venv\Scripts\activate

# macOS/Linux
source venv/bin/activate

Install dependencies:

pip install -r requirements.txt

🔐 Environment Variables

Create a .env file in the project root:

GOOGLE_API_KEY="your_api_key_here"

Do not commit your .env file or API keys to GitHub.

▶️ Run Locally

streamlit run app.py

Then open the local URL shown in the terminal.

🐳 Docker Usage

Build the Docker image:

docker build -t youtube-rag-chatbot .

Run the container:

docker run -p 8501:8501 --env-file .env youtube-rag-chatbot

Open in browser:

http://localhost:8501

☸️ Kubernetes Deployment

Create a Kubernetes secret for the Gemini API key:

kubectl create secret generic youtube-rag-secrets --from-literal=GOOGLE_API_KEY="your_api_key_here"

Apply the Kubernetes manifests:

kubectl apply -f k8s/deployment.yaml
kubectl apply -f k8s/service.yaml

Check running pods and services:

kubectl get pods
kubectl get svc

⚙️ CI/CD

This project includes a GitHub Actions workflow that runs on pushes and pull requests to the main branch.

The workflow:

• Checks out the repository
• Sets up Python 3.11
• Installs project dependencies
• Validates Python syntax for app.py and main.py

💡 Usage

1. Open the application.
2. Paste a YouTube video URL.
3. Enter a question about the video.
4. Click Get Answer.
5. The chatbot retrieves relevant transcript context and generates an answer.

Example questions:

• What is this video about?
• Summarize the video.
• What are the main points?
• What does the speaker recommend?

🧠 Skills Demonstrated

• Retrieval-Augmented Generation
• LLM application development
• LangChain pipeline design
• Vector search with FAISS
• Google Gemini API integration
• Streamlit application development
• Docker containerization
• Kubernetes deployment
• GitHub Actions CI
• Cloud deployment

🔗 Links

• 🌐 Live App: YouTube RAG Chatbot
• 💻 GitHub Repository: DevaChaudhari/YouTube-RAG-Chatbot

👨‍💻 Author

Deva Chaudhari

GitHub: @DevaChaudhari

📄 License

This project is open source and available under the MIT License.