Founder of SayMake · Human–AI learning systems · Open standards, agent infrastructure, and AI safety
I build AI-supported, project-based learning that helps young people turn what they care about into real-world capability—the skills, earned confidence, and agency to shape what happens next.
What kids build changes who they become.
AI should scaffold thinking, not surrender it.
| Project | Purpose |
|---|---|
| SayMake | An AI-supported maker studio where kids turn personal interests and real problems into projects they can build, test, explain, and improve. |
| OMBS | The Open Making and Building Standard: an openly licensed, machine-readable framework for making the learning behind real projects visible. |
| SPEAR | Defense-in-depth security middleware for LLM and agent pipelines, including prompt-injection defense, provenance, tool mediation, and output protection. |
- Does an AI system leave the person more capable after assistance?
- Which decisions should remain with the learner, and which should AI support?
- How can we distinguish an impressive artifact from demonstrated human capability?
- How should AI assistance adapt, fade, and preserve meaningful agency?
- How can we prove that lower cost and greater automation did not reduce quality?
- How can project evidence become portable across tools, schools, and contexts?
- Claude Starter Kit — reusable agents, skills, hooks, and MCP configurations for practical AI-assisted work
- ContextStellar — context observability and prompt-efficiency tooling for LLM applications
- SessionStellar — evaluation and scoring infrastructure for agent orchestration quality
- AI Skills — reusable workflows for structured thinking, project hygiene, and AI-agent operation
Before founding SayMake, I spent two decades building software, data, and machine-learning systems at Apple, Microsoft, and Salesforce.
That experience shaped how I approach AI learning systems: as products that require strong architecture, observability, evaluation, safety, privacy, and evidence—not merely compelling model output.
My work sits at the intersection of:
- Human–AI collaboration
- AI-supported project-based learning
- Learner agency and competency development
- AI evaluation and verification
- Agent harnesses and context engineering
- Safety, privacy, and local-first systems
- Open standards for capability evidence