Built for ambiguity, tuned for clarity.
I design AI products and dense workflows by tuning three things at once: system logic, visual precision, and live prototyping speed.
Drag a parameter. Watch the profile recompose.
Systems first. Interfaces second.
Deeply influenced by CMU MHCI. I do not just skin components; I architect multi-step discovery flows and tame inference latency. I speak the same language as data models.
What I run on. What I refuse.
Senior taste is not just what you add. It is what you deliberately remove: empty automation, decorative AI, and static case-study bureaucracy.
What I Run On
I treat AI models, edge cases, and constraints as design materials, not decoration layered on top of a static interface.
AI should augment people, not erase their agency. The user should always understand, steer, and recover.
I use Codex and Cursor to validate native HTML behaviors quickly, so system logic gets tested before it becomes theater.
What I Deprecate
No soulless component stacking, no AI chatbot pasted onto a product without a workflow reason.
I reject cold automation that strips creators and operators of control at the exact moment judgment matters most.
Less timeline bookkeeping. More evidence that the system works, the logic scales, and the product has a point of view.