project / March 2026

LLM assistant for a simulated smart home

A full-stack prototype combining an assistant backend, semantic memory, Home Assistant integration, UI work, and fall-detection ideas.

LLMFastAPIReactHome AssistantPrototype

This project was a full-stack assistant prototype for a simulated smart-home environment. It combined an assistant backend, semantic memory, Home Assistant integration, a React interface, voice and service modules, and ideas around fall detection.

It is not the most security-focused project in my portfolio, but it shows how I build across layers. The backend has modules for intent classification, context composition, semantic retrieval, services, text-to-speech, weather, alarms, and Home Assistant control. The frontend makes the system usable instead of leaving it as a set of scripts.

The security angle is the one I care about now: once a system can act in a home environment, permissions, logging, input boundaries, and failure behavior become serious. A helpful assistant that can trigger services needs careful limits.

For internship discussions, I use this project as evidence of prototyping speed, backend/frontend integration, and the habit of thinking about safety when software touches the physical world.