Built during HackUTA 7, CompanionAI connected a React microphone interface to Gemini-assisted interpretation, a Flask service, and a Supabase-backed command queue. The Raspberry Pi continuously polled for commands to turn speech into physical behavior.
Engineering contributions
- Integrated audio transcription and intelligent command interpretation
- Designed a PostgreSQL-backed real-time command polling flow
- Connected the web interface to Raspberry Pi hardware through REST APIs
- Built reusable React components for microphone interactions
System path
01Microphone
02React UI
03Flask API
04Supabase
05Raspberry Pi