Built a Chrome Extension that helps users summarize content and extract key concepts from:

• YouTube videos • Web browser page text

All directly from the browser using a Retrieval-Augmented Generation (RAG) pipeline.

Tech Stack Used

-> Frontend • HTML, JavaScript • Chrome Extension APIs (Manifest V3) -> Backend • Flask (Python) -> AI / LLM Stack • LangChain • LLM: HuggingFace – Qwen/Qwen2.5-1.5B-Instruct • Embeddings: HuggingFace – BAAI/bge-small-en-v1.5

RAG Indexing Pipeline

-> 1. Document Loaders • YouTube: YouTubeTranscriptApi • Web pages: WebBaseLoader -> 2. Text Splitting • RecursiveCharacterTextSplitter -> 3. Vector Store • FAISS for similarity search

• Implemented an end-to-end RAG workflow (load → split → embed → retrieve → generate) • Integrated LangChain with a Chrome Extension frontend • Built a clean separation between browser-based UI and Python backend • Used open-source LLMs and embeddings instead of paid APIs