I integrate production AI workflows into real applications, including LLM orchestration, real-time speech-to-text, text-to-speech, voice agents, structured outputs, provider fallbacks and auditable usage controls. I develop Mendelio's AI and voice platform across Next.js/React, TypeScript, Cloudflare Workers, Supabase/PostgreSQL, Python services and GPU-backed TTS infrastructure. I can design a new integration or rescue an unreliable one, improve latency and observability, protect billing and data boundaries, and deliver tested deployment-ready changes rather than a fragile demo.