Working with large documents often requires searching through pages to find relevant information. DocChat AI solves this by enabling natural language queries against uploaded documents, providing contextual answers with source references.
Implemented a Retrieval-Augmented Generation (RAG) pipeline using Google Gemini API. Documents are parsed, chunked, and indexed to enable intelligent context-aware responses. Built on Supabase for secure authentication and data persistence.
Built the frontend with React and TypeScript, integrated PDF.js for document parsing and in-browser viewing. Used Supabase Auth for secure user management and Supabase Edge Functions for server-side RAG processing with Google Gemini API. Styled with Tailwind CSS for a clean, responsive interface.
Built the chat interface and document management UI
Type-safe codebase ensuring reliable document processing
Powered the RAG-based conversational AI engine
Authentication, database, and Edge Functions for server-side logic
Client-side document parsing and in-browser PDF viewing
Responsive, utility-first styling for the chat interface