The assistant above is powered by a FastAPI backend that integrates with a large language model via Groq API. Responses are grounded using a controlled persona and knowledge bundle, ensuring accurate and relevant answers. The system is containerised and deployed to Render, with API credentials securely managed through environment variables.
The frontend is built with React and Vite, deployed on Vercel with Git-based continuous deployment. This page embeds the assistant securely with CORS restricted to approved domains. The architecture mirrors how I would structure a real-world product integration: typed backend API, separated frontend interface, and secure configuration management.