I build things where AI, product, research, accessibility, and human behavior collide.
Most of my work starts the same way: I notice something that isn't working quite right, pull at the thread, and eventually build the thing I wish had existed.
Right now I'm experimenting with:
- human-centered AI evaluation
- agent systems that know when not to act
- tools for neurodivergent workflows
- turning research and judgment into reusable systems
- occasionally making something weird just because I want to see if it works
Carousel Studio — live demo A low-spoon Instagram carousel builder: eight reusable format anatomies, print-clean 1080×1350 exports, contrast checked to WCAG 2.2 AA — and everything runs locally, so your photos never leave your device.
Career Pivot Navigator AI-assisted career exploration for neurodivergent and burned-out professionals. Realistic exits over toxic positivity, tactical plans over vague inspiration.
VisionPT / StabilityLabs A high-fidelity spatial computing prototype for EDS physical therapy on Apple Vision Pro: isometric protocols, built-in joint-safety guardrails, glassmorphic interface. A prototype, not a production app.
90/10 Agent Benchmark Testing whether AI systems can actually reduce human cognitive load instead of creating more work. Still early — this is where the benchmark is taking shape.
I investigate what breaks. Then I design what should have existed.
That might turn into a product, a prototype, a research system, a framework, an article, or something I didn't have a name for when I started.
UX. AI evaluation. Research systems. Accessibility. Agent behavior. Neurodivergent product design. Why a supposedly "simple" workflow is making people miserable.
Or ask me what I'm building this week.