Projects
Examples of turning messy context, repeated work, and agent workflows into systems with review gates, operating docs, and visible outputs.
Knowledge OS shows how raw signals, source review, agent work, durable memory, and reviewable artifacts become trusted company knowledge instead of context sprawl.
Jules is the internal operating lab for governed AI workflows: decision support, execution, verification, publishing prep, reporting, and handoffs.
Decision advisory workflow
A decision pattern for bringing multiple expert perspectives into one clear recommendation before work begins: domain context, critical review, implementation judgment, and a final human call.
Ava Attendance Officer
Hackathon 2025A multilingual school attendance agent. Connects a teacher dashboard, hall-pass iPad, policy backend, and an OpenAI Realtime voice agent into one working loop. Ava detects overdue passes, evaluates policy thresholds, and calls guardians in English or Spanish.
Operations
Local agent infrastructure, service supervision, launchd jobs, Cloudflare deploy paths, credential flow, and the daily plumbing that keeps an AI business running.
Event setup, publishing, scheduling, RSVP flow, follow-up materials, and the coordination work behind public demos and operating-system launches.
Small tools that remove friction: Whisper repetition cleanup, Plane sync, markdown workflows, credential helpers, and automation scripts for recurring work.