I turn ambiguous ideas into AI products people can use, inspect, and trust.

Graded Chinese listening from real stories, with synced pinyin and translation. Web, iOS, and podcasts.

• Shipped the learning loop so stories, audio, and vocabulary match each learner's HSK level.
• Built the web and iOS product people use to listen, look up words, and come back the next day.

An AI coach for amateur athletes. It reads their training data and writes plans they can understand and change. Chinese beta.

• I own the agent: tools over activity data, a coach that proposes and revises plans, and the evaluation harness that checks those suggestions.
• I also built most of the product the agent sits in: activity import, planning, teams, and the workspace; beta-tested with 30 active runners across two university running groups.

Let people and external agents create, iterate, and deploy applications through an interface they can trust.

• Built Atoms CLI from 0→1, giving coding agents a direct path to create, inspect, validate, and deploy applications.
• Defined a full-stack data workspace so non-technical builders could inspect and update app data without SQL.
• Shaped the platform across product strategy, interaction design, and engineering, documented through 15 RFCs and PRDs.

A cross-platform timer that makes breaks harder to skip while keeping its rules transparent.

• Designed a work-and-rest product for Windows, macOS, and Linux around transparent rules and explicit user control.
• Grew it over six years into a multilingual open-source product with an international contributor community.

An AI-assisted paper-cutting tool that helps creators work with cultural symbolism, not shape alone.

• Designed a system that connects visual form with cultural meaning, then evaluated it in a 16-person study.
• Published the work as a full paper at ACM CHI 2025.

A computer-vision defect-detection system that learned from normal production samples and ran in real time on low-cost edge hardware.

• Co-led a deployed quality-inspection system, pairing normal-only anomaly detection with a dashboard shaped by factory-floor observation.
• Optimized the vision pipeline for real-time inference on approximately $40 hardware and deployed it across 20 loom stations, with operators clearing defects and managers monitoring production status.