I don't prompt AI. I architect with it.
VP of Platform and de facto AI lead at a B2B SaaS company — where I rebuilt a 20-person operation down to a 4-person team running at the same output, using AI systems I designed and built myself.
My background is design and product. I think in systems, build under constraints, and document everything so it survives without me.
- Agentic workflows — multi-source ingestion, priority logic, structured outputs, and feedback loops that make systems smarter over time
- Reference architectures — shared knowledge layers that multiple workflows depend on; change once, propagate everywhere
- Operational AI tools — weekly revenue pipelines, executive engagement copilots, client health monitoring, data enrichment at scale
- Pre-funding foundations — architecting the AI operating layer before the dev resources arrive, so when v2.0 starts, the system is already proven
| Repo | What it is |
|---|---|
personal-health-os |
A three-skill AI system for tracking training, food, and recovery — built in Claude |
concierge-live-status |
Live operational dashboard built on top of a platform that couldn't give me what I needed. Reads 11 client trackers in real time via Google Drive MCP |
persona-builder |
A skill that interviews you and builds three MD files teaching Claude who you are, how you work, and how you sound. One conversation, done in under an hour |
doc-formatter |
A Claude skill that turns meeting notes, transcripts, and brain dumps into a styled HTML artifact and Google Doc in one command |
Building the operational AI foundation for a B2B SaaS marketplace — designing agentic workflow systems, shaping the v2.0 product architecture, and creating the recurring revenue predictability that makes funding close.