Notes

How the systems actually work.

Write-ups on the agents, voice pipelines and products I build: the architecture, the trade-offs and what I would do differently.

  1. · 3 min readAnatomy of a real-time AI voice agentWhat it takes for an AI to hold a phone call: voice activity detection, streaming speech-to-text, a grounded LLM, fast speech synthesis, and a clean hand-off to a human.
  2. · 5 min readAskora: an AI workstation that runs from a USB driveChat, web search, image and video generation, an in-browser video editor, a medical report reader and a 3D anatomy atlas, all running offline from one portable folder on a 4 GB laptop GPU.
  3. · 3 min readHow this portfolio worksA talking intro video, a 3D mini-me you can ask questions out loud, narrated demo videos, and hosting that uses about 70 MB of RAM. Here is what is behind each piece.
  4. · 3 min readRunning an AI agent team from TelegramHow Hermes and OpenClaw split projects, ship code under their own GitHub accounts and run marketing for three brands, with a human approving everything public.