Back to all projects
AI & ML2025
Personalized RAG Chatbot
A privacy-first Retrieval-Augmented Generation chatbot built with a locally hosted Mistral-7B-Instruct model and GPU-accelerated inference.
The Problem
Securely querying massive legal documents locally to protect privacy while maintaining high-quality answers.
The Approach
Implemented a keyword-based RAG pipeline grounding responses in a curated knowledge base via dynamic context injection.
My Role
Solo AI Engineer. Built the full pipeline—from the Flask backend handling retrieval and prompt construction to the custom web interface exposed via Cloudflare.