Multi-agent AI systems that run real operations.
I orchestrate frontier and self-hosted models into agent fleets that ship code, run infrastructure, and execute operations. Nothing goes out on one model's say-so.
Z.ai
Moonshot
Sixteen years, my own shop.
Computer scientist by training, B.S. Computer Science with a Business Administration minor from Southern Oregon University, 2010, and an independent technology founder ever since. I have run my own S-corporation for sixteen years, and over the last few I rebuilt the entire operation around applied large language models.
I own the whole stack and ship end to end. Nothing I build is a demo. It runs my own business, and when it breaks it costs me. That is a harder test than any client review.
One pass is a coin flip.
A model checking its own work agrees with itself, confidently, every time. So the reviewer is always a different lineage: different training, different failure modes. That much is table stakes.
What isn't: I don't stop at one exchange. Models from different lineages argue in a single shared thread, for as many rounds as the disagreement takes, until it resolves instead of averaging out.
It matters more than it sounds. In a recent run the first answer was wrong. The correct one did not exist until the fourth round. Four separate errors surfaced along the way, and not one of them was caught by the model that made it.
A single adversarial pass would have shipped round one, with citations attached.
What I actually run.
Orchestration
Coordinating frontier models with self-hosted open-weight LLMs over MCP. Agentic coding is the daily driver, not an experiment.
Hermes · localLocal inference
Real compute I own and run, not a rented endpoint. 446 GiB of open-weight models resident on the box, two of them loaded at once: one to reason, one to answer instantly. Speech is local too: an RTX 5080 runs Whisper large-v3 for dictation and video transcription, so none of it leaves the building either.
The large model reasons. The small one answers instantly. Both stay resident at the same time, because the memory is mine and nobody meters it.
Languages
I ship in whatever the problem needs, on a CS foundation that goes all the way down.
Infrastructure
Linux servers, plus a 2021 workstation I rebuilt into one. When it came out of daily service I wiped Windows off it, put Linux on, and gave an autonomous agent real work to do there. That is where the self-hosted side of this came from, and I learned it by operating it rather than reading about it. It runs my private Git and automation now. My own Git, my own servers, my own models.
Model work
Prompt and context engineering, evaluation, adversarial red-teaming, RAG, and technical and code data generation.
AI engineering, model evaluation, applied LLM work.
Remote, or Phoenix hybrid. You get someone who already lives inside these systems at production stakes every day, not someone ramping into them.