I build practical tools that make advanced AI easier to operate, understand, and trust.
| Project | What it demonstrates |
|---|---|
| Carina AI Command Center | A polished multi-agent operations interface with explicit prototype boundaries. |
| Maya Orchestration Engine | Risk-aware routing, native macOS feedback, voice control, and a private authenticated relay. |
| Codex Click Guide | An accessible native overlay that turns assistant handoffs into visual and spoken click targets. |
- Human-centered interfaces for complex AI workflows
- Local-first automation using Swift, Python, Shortcuts, and native macOS tools
- Agent coordination with visible state, bounded permissions, and approval gates
- React and TypeScript prototypes that turn ambitious system ideas into testable products
- Build real, runnable software and describe prototypes honestly.
- Keep sensitive credentials and private state out of source control.
- Make important system behavior visible to the person operating it.
- Treat accessibility, safety, and verification as product features.
I am connecting the Carina interface, Maya's orchestration layer, and Click Guide's accessible return loop into one coherent human-in-the-loop system.
Python · Swift / AppKit · TypeScript / React · macOS Shortcuts · Tailscale · GitHub


