Ken Molina — AI-Orchestration Engineer
I build multi-agent AI systems that run themselves —
and prove they're working.
Not demos. Production systems where a supervisor plans, workers execute in parallel, and an adversarial validator has to sign off before anything ships. I design the agent topology, operate it on my own homelab, and keep the full run trace so every result is reproducible.
§ 01 / EXHIBITS
Selected systems
02 shown · each shipped & running 01 jn-bot running A read-only audit agent that reads one roofing-CRM account end-to-end, walks five hygiene questions, and surfaces evidence-cited risks a manager can act on — and throws out its own findings when it can't ground them. open exhibit → 02 ai-radar running A cron-fired agent that sweeps the AI landscape every morning through six parallel lenses, writes a project-linked digest into my vault, and publishes a fail-closed, headlines-only snapshot to this page. open exhibit →