Retrieval-Augmented Generation System

RAG Chatbot

A RAG-based chatbot with LangChain, Llama 3, and Chroma vector DB - deployed as a Chrome extension for instant access on any webpage.

RAGLangChainLlama 3Chroma DBOpenAI EmbeddingsFlaskReact
GitHub Repository

Overview

This project implements a full Retrieval-Augmented Generation (RAG) pipeline that allows users to chat with custom knowledge bases. It is deployed as a Chrome extension, making it accessible from any webpage without leaving the browser.

Architecture

  • Web scraping with BeautifulSoup for knowledge base ingestion
  • Text preprocessing via stemming and lemmatization for NLP readiness
  • Vector storage in Chroma DB using OpenAI Text-Embedding-3-Large model
  • RAG pipeline built with LangChain stuff document chain and Llama 3
  • Flask backend exposing RESTful endpoints for the extension
  • React frontend embedded in the Chrome extension for seamless UX
  • Few-shot and chain-of-thought prompting for improved answer quality