Product-minded builder at UCSD Math-CS.
I turn messy workflows into small systems with visible state.
AI product loops 路 frontend systems 路 instrument automation 路 product taste
I like software that shows its work: what changed, where the state lives, what the AI is allowed to do, and how a user can replay or trust the result.
Currently I am closest to the intersection of product judgment, frontend experience, and system implementation. I care less about adding more features and more about making the core loop feel clear.
| Project | What it is | Small proof |
|---|---|---|
| Tiny Stories / RPG_Demo | AI narrative runtime: seed -> play -> inspect -> replay. | React + FastAPI, typed contracts, reviewer mode, release gate, public demo. |
| Auto Load-Off Test | Desktop tool for repeated AWG/oscilloscope sweep measurement work. | Python + PyVISA/SCPI, 30 hardware-free tests, CI, docs, about 75% less setup/result organization effort. |
- Start from a real workflow, not a feature list.
- Make state visible before making the UI flashy.
- Keep AI inside clear boundaries: context, permissions, logs, replay.
- Ship small demos that can be inspected by another person.
Building portfolio-ready systems around AI product workflows, reliable automation, and interactive tools that make complex systems feel easier to control.
Python 路 TypeScript 路 React 路 FastAPI 路 SQLite 路 PyVISA/SCPI 路 LLM workflows