CS + physics at Cal State Long Beach (biology minor), BUILD Scholar working toward MD/PhD training. I direct AI agents to build research software, games and tools, and I decide what ships. Checks come before claims: runnable demos, public source where possible, and explicit limits.
- Hamon — my studio: tools for AI, experiences for people.
- Hamon Engine (private source) — a Rust + wgpu engine built for AI agents to drive: typed ECS commands, replay and agent control. Current work is a humanoid character creator whose quality is held by numeric gates (seams, penetration, stretch, gait) rather than eyeballed renders — species bodies, tails and ears with spring physics, and strand fur.
- Wildside — enter the living worlds of the places around you. Playable in the browser.
| Project | What it is |
|---|---|
| TowerOps | Synthetic ATC conflict-awareness and bounded-agent research demo; not operational ATC. |
| CaptureSuite | Windows-first multimodal research capture: daemon, versioned worker plugins, sealed sessions, PySide6 UI. |
| NULLGARDEN | Executable negative controls for scientific ML — does that AUC survive when the signal is removed? |
| Experimental Earth | Local workbench for investigating water, shade and terrain changes to a real place. Early. |
| TestPilot | Agent that writes, runs and repairs regression tests for a diff (Nebius × NVIDIA hackathon). |
| household-ledger-mcp | Typed TypeScript MCP server for household income, expenses and budgets. |
| CanvasPilot | MCP server + CLI for Canvas LMS that rides your existing browser session. |
| RecallWeave | Browser-only learning toolkit with a transparent Bayesian learner; no learning-gain claim. |
Small prototypes (Oct 2026): forkscope · worldlaw · equivia · masks · shadow-stage · common-test · crosswire · quiet-mesh · assumption-arcade · pausepoint · loopkind · ludolingua
Also: workflow-checks · multi-agent-fanout-engine · uma-sim · Spire of Octaves (my fiction world)
Gait and rehabilitation, behavioral neurobiology, protein molecular dynamics, and collaborative surgical-ergonomics work. Research contributions are collaborative; synthetic demos are not evidence of clinical validity or savings on someone else's data.



