About
Executive coach and agentic systems architect. I build governance infrastructure for AI systems and coach technical leaders through the decisions that shape them.
The arc
I started in human performance — strength and conditioning coaching, where I learned that outcomes follow process discipline. A training program works when you can measure progress, identify failure modes early, and adjust before the athlete stalls. That same discipline transfers to AI systems: governance is the training program for your agent fleet.
Over a decade, I moved from coaching individual performance to coaching technical leaders and building the infrastructure they depend on. The throughline is the same: assess constraints before prescribing solutions, build safety infrastructure before features, and manage change carefully.
What I do
- Executive coaching for technical leaders — I help founders, platform leads, and AI team leads make better decisions about AI architecture, governance, and team structure. The coaching is outcome-focused: we define what success looks like, identify what breaks in production, and build the infrastructure that prevents it.
- AI governance infrastructure — I built and operate HUMMBL, an open-source governance primitives library for agentic AI systems. 34 primitives, 2,463 tests, zero runtime dependencies, published on PyPI.
- Production multi-agent systems — I operate a governed multi-agent AI platform running daily in production. This is not a demo — it is live infrastructure that I depend on and maintain. The governance primitives are public (GitHub, PyPI); the platform itself is internal.
Selected work
- hummbl-governance — Runtime governance primitives (kill switch, circuit breaker, delegation tokens, audit log, capability fence, 30 more). Published on PyPI, Apache 2.0, zero third-party Core dependencies. Verify on GitHub →
- hummbl-governance — Internal governed multi-agent AI OS. The proving ground for hummbl-governance primitives — daily production operation across 5+ AI models. Source is private; governance primitives extracted from it are public.
- Base120 — Reasoning framework with 120 operational models across 6 transformation families. Deterministic governance substrate for system design and validation.
- The Governance Tuple — Published research on representing governed decisions as a triple (Contract, Delegation Context Token, Evidence). Available on Zenodo.
Skills
| Domain | Specifics |
|---|---|
| Languages | Python, TypeScript, JavaScript |
| AI / ML | LLM evaluation, agent orchestration, multi-agent coordination, RAG systems, prompt engineering, MCP, model drift diagnosis |
| Infrastructure | CI/CD, Cloudflare Workers/Pages, FastAPI, Next.js, automated testing (unit, integration, e2e, chaos, security), observability |
| Governance | NIST AI RMF, ISO 42001, EU AI Act, kill switch / circuit breaker architecture, append-only audit logging, delegation tokens, capability fences |
| Coaching | Executive coaching, technical coaching, outcome framing, pre-mortem facilitation, cross-functional team leadership |
Location
Atlanta, GA. Remote (US).
Want to talk?
Book a 30-min intro →