I build voice and communications systems with applied AI. My work covers telephony, transcription, data pipelines, agents, and the infrastructure around them.
At AnSer, I build full-stack voice agents with LiveKit across telephony, real-time audio, speech-to-text, LLM and tool orchestration, retrieval, text-to-speech, evaluation, product interfaces, and production infrastructure.
I also build the AI-assisted QA pipeline around our Kafka and CDC streams. Calls move through capture, transcription, structured extraction, applicability and gradability rules, automated evaluation, adjudication, reporting, and supervisor review.
That work includes shared Fastify and APISIX APIs, PostgreSQL and pgvector, Next.js and React products, and Dockerized delivery on Linux and Proxmox.
I like the part after the demo: noisy audio, latency, incomplete context, and figuring out what happened when the system gets something wrong.
I spend most of my time in Python and TypeScript, but the work regularly pulls in the tools below.
Languages & product
AI, voice & data
Platform & delivery
