The systems
Four production systems, each built end to end and operated for real: designed, shipped, monitored, and kept alive by one person.
An AI operations control center built on Claude: autonomous agents doing real work, always-on monitors, push alerts, morning briefings, and a live dashboard running 20 services 24/7.
An NFL and college-football data platform: normalized schedules and betting odds, idempotent ETL, weekly client deliverables. Rebuilt this year behind a 242-test parity suite proving old and new outputs identical.
A payments SaaS built from zero as co-founder and CTO: checkout flows, payment-provider integrations, and three years of owning architecture, releases, and reliability.
A lead-working tool for asset-recovery investigators: 3.3 million California unclaimed-property claimant rollups, a status workflow, full-record drill-downs, and a deterministic compliance layer behind the outreach.
ALSO OPERATING: UNKND Labs, a physical-products business (design → 3D-print → sell): 252 sales, 4.9★ on Etsy, with an automated production pipeline.
How I work
Production first
A demo is the first 20%. The work I care about is the rest: error pipelines, health monitoring, safe restarts, recovery paths. My systems are built to run every day under real load.
AI with guardrails
Agents get scoped tool access, lifecycle management, and verification gates. The lesson from running Claude agents daily is that autonomy is earned through constraints.
Owner, end to end
Three years as a CTO taught me to own the whole loop: architecture, code, deployment, incident, fix. There is no part of the stack I hand off.
Get in touch
If you need someone who can build an AI system and actually run it, I'm in Los Angeles and I answer fast.