Professional experience
Vision On
AI Image Research Intern · July 2026 – Present
- Built production benchmarks to evaluate 20+ AI image models for commercial retouching, testing model strengths and failure modes across 10 production scenarios with senior retouchers.
- Built a model-selection taxonomy mapping production tasks to model capabilities and failure modes; adopted into the studio’s AI retouching pipeline.
- Shipped 100+ client-facing AI assets and 10+ advertising videos, turning model experiments into repeatable production workflows.
Ali Lin Lab LLC
Founder · Design Engineer · June 2026 – Present
- Built ACUBOT with Expo and React Native, creating a bilingual learning system covering 136 acupuncture points and 4,000+ clinical cases; designed the search, content architecture, and interaction system. Currently in closed beta.
- Rebuilt a Texas acupuncture clinic’s website from WordPress to a static Vercel deployment, redesigning the information architecture around five patient-centered service pathways.
HFunds
AI Creative Engineer Intern · Dec 2024 – Apr 2025
- Built an end-to-end AI content pipeline with OpenAI and Hugging Face APIs, cutting per-asset turnaround by roughly 30 percent across the creative team.
- Bridged ML capabilities and product design, translating model behavior into production-ready creative workflows under tight launch timelines.
Selected projects
WebGL · GLSL · TypeScript · React
- Built a real-time film physics engine in WebGL and GLSL, modeling halation, grain structure, and film response for AI-generated video entirely in the browser GPU.
- Calibrated film profiles against original CineStill 800T and Portra 400 scans, using measured film behavior rather than published LUTs.
TypeScript · React · fal.ai API
- Built an interactive tool to instrument diffusion inference end-to-end, exposing latency, convergence, sampling behavior, and queue dynamics hidden behind hosted APIs.
- Measured 47 inference calls and isolated the dominant source of latency variance: queue time (277 ms std) versus inference (9.9 ms std), with raw measurements shipped alongside the report.
Claude Vision · GLSL · TypeScript
- Built a vision-to-shader pipeline that extracts material parameters from fabric images with Claude Vision and drives a real-time GLSL material simulation.
- Added validation, rate limiting, error handling, and shader export to make the prototype production-ready.
Skills
- AI
- Claude Code, Cursor, Anthropic API, OpenAI, Hugging Face, fal.ai, Replicate, diffusion evaluation, VLM pipelines
- Languages
- TypeScript, JavaScript, Python, GLSL, Solidity
- Graphics
- WebGL2, GLSL, Three.js, React Three Fiber, shader authoring, color science, WebCodecs
- Frontend
- React, Next.js, semantic HTML, modern CSS, Tailwind, Expo, React Native
- Tools
- Git, Weavy, Figma, Photoshop, Lightroom, Capture One
- Photography
- Medium-format film & digital. Four years of editorial work across three NYU publications; 100+ commissioned shoots.