Technical Program Manager at eBay, reporting to a VP of Engineering. My peers were directors and engineering managers. My day job was getting them to agree on trade-offs across large engineering programs. I do that job for AI portfolios now, and I ship the systems myself. Founder of baxie. Not an advisor. Not an automation shop.
Setting the direction is one job. Shipping against it is another. Almost every fractional AI offer in 2026 picks one and subcontracts the other, which is exactly where the dependencies go unmapped and the work stalls. This profile carries both: program execution on large engineering organizations at eBay, named frameworks with their origins cited so your CFO can check the source, and patterns running in production today.
Technical Program Manager at eBay, reporting to a VP of Engineering, before AI was production-ready. I map AI dependencies the way I mapped engineering programs: owners, integration points, risk, downstream impact. The layer almost every AI consultant skips, because most of them have never run a program.
Three-Horizon AI portfolio (McKinsey, 1999). Leadership-Lab-Crowd adoption (Ethan Mollick, Wharton, May 2025). Agentic pattern catalog: ReAct, tool-using, multi-step, sub-agents, self-correction. Origins cited so your CFO can quote the source and your board can audit the logic.
Vision extraction on architectural plan sets behind a review gate enforced in the data layer. Multi-model routing tuned for cost and latency. A deterministic verifier that flags the model's outliers without overwriting them. A closed scope catalog that stops scope-name hallucination. A markup engine that throws rather than return a nonsense number. Running in baxie production.
Org redesign and the build crew, from one operator. Not a deck. Not a referral.
Technical Program Manager on eBay's Product Strategy and Operations team, working across Core Product and Core Technology, its two largest engineering organizations. Mechanical Engineering foundation from SFSU. Builder hands that now run a production AI stack.
Technical Program Manager on Product Strategy and Operations. Mapped engineering program dependencies across owners, integration points, risk, downstream impact. The discipline that translates directly to AI portfolio rollouts that land.
Bachelor's in Mechanical Engineering. Systems thinking, tolerance analysis, root-cause discipline. Same posture I bring to AI systems that have to behave in production.
Baxie in paid beta. Three engineers' worth of work shipped by one founder on a shared knowledge base and a production AI stack. The proof is current, not historical.
Fractional Head of AI Transformation work. Org redesign, production AI build, workforce training, from one operator. The service ladder runs from AI Portfolio Allocation Reviews through shared-knowledge-base buildouts, scoped production builds, training, and AI System Continuity as the maintenance retainer.
Three-Horizon AI portfolio allocation across H1 efficiency, H2 capability, H3 transformation. What to fund this quarter, what to kill, what your board needs to see.
AI-ready knowledge bases modeled on edwin-os. Versioned. Searchable. Both your team and your AI read from it. Sold as a scoped build, not a deck.
Vision extraction with human review gates, multi-model routing, deterministic verifiers, and prompt-injection hardening on document intake. Patterns I run in baxie production, applied to your problem.
Same architecture I sell to clients. The shared knowledge base is the substrate. The work compounds because the context is one layer.
AI-native Margin OS for residential general contractors. Paid beta. Three engineers' worth of work, shipped by one founder running off a shared knowledge base and a production AI stack.
Fractional Head of AI Transformation engagements for operators and founders. Org redesign, production AI build, workforce training, from one operator.
Engagements that violate these don't work. So I don't take them.
If you haven't shipped a pilot, the right work is the AI Program Diagnostic, not a 6-month retainer. The retainer pays off when there is real production behavior to redesign around. Otherwise we are designing in a vacuum.
Every pattern I sell is one I run in Baxie today. Multi-agent orchestration, multi-model routing, vision pipelines, closed-loop learning, the shared knowledge base. If I don't run it in baxie, I won't sell it to you.
Model-agnostic and tool-agnostic. I make money from your engagement. No referral fees, no reseller margins, no vendor partnerships. The stack we pick fits your work.
Every engagement ends with your team running the eval suite, playbook, and on-call. Not a permanent dependency. If your team can't take it after handoff, the engagement isn't done.
Book the call. 30 minutes. I'll tell you in the first conversation whether I'm the right operator for what you're trying to ship.