Avicenna / Lumina / SOMA
Full-Stack Web Developer at AcceliDx
Voice-first, camera-aware smart-mirror AI health coach. Users are recognized by face and voice and coached through natural conversation, with touch reserved for sensitive input. Built as two deployable units — a Next.js 16 / React 19 frontend on Vercel and a unified FastAPI biometric backend (Dockerized, on Hetzner) — with real-time conversational voice over the OpenAI Realtime API, face + voice authentication, and Supabase-backed identity, transcripts, and usage accounting.