Product manager who builds — with AI. I'm a PM on Omnissa's Freestyle Orchestrator team. The fastest way I understand a problem is to build something that solves it, so most of what's here is real, working software — not slideware. I use AI tooling to build it, and I've shipped enough projects to know the logistics and the gotchas.
What I build spans AI agents, developer tools, and privacy-first apps — plus a steady stream of things for personal interest and everyday productivity.
🔭 Exploring how automation and AI agents reshape the digital workspace
🧰 I build with AI tooling — across whatever stack the problem needs (Go, Python, Swift, TypeScript)
📍 Atlanta, GA
| Project | What it is |
|---|---|
| ws1-uem-agent | An agent-shaped CLI + Claude skill that lets an AI safely operate enterprise device management — browser-approval for destructive ops, blast-radius gating, hash-chained audit log. |
| htmd | A Markdown ↔ HTML bridge for AI agents. Agents emit ~150 tokens of YAML; humans get a polished interactive page with an export button that closes the loop. 7–34× cheaper than hand-written HTML. |
| workspace-one-uem-mcp | An MCP server exposing 25+ Workspace ONE UEM operations to any AI assistant. |
| Freestyle-UUID-Resolver | A Chrome extension that resolves Workspace ONE UEM UUIDs into human-readable entities from the right-click menu. |
What I'm actively building right now — most are private or pre-launch, so no links yet:
- aloud — local, private voice dictation for macOS; on-device speech models, nothing leaves the machine (in evaluation for company-wide use).
- clone-a-claw — a marketplace for AI-agent personas, with a safety scanner that strips secrets before anything ships → cloneaclaw.com.
- pm-toolkit — a context-grounded PM copilot: Claude skills plus a personal "context pack" (Profile / Projects / Counterparts / Vision).
- personal-finance-tracker — a self-hosted, local-AI "Mint replacement" for me and my family.
- AlgoTrade — a personal algorithmic futures trading system; backtested and now paper-trading.
- Start from the problem — I dig into the real user/business problem first; building is how I pressure-test the solution.
- Outcomes over output — I measure what a thing changes once it ships (adoption, before/after, token receipts), not how much I produced.
- Human-in-the-loop by default — agents act, people approve; the design work is the approval moment.
- Trust is a feature — fail-closed defaults, privacy by design, audit trails: what makes powerful tooling adoptable.
- Patterns travel — I port good ideas across domains (OpenClaw's identity files → a PM context pack → a persona marketplace).