The other side of building
AI speeds up building, but every product needs marketing, design, compliance, and care. Quality UI signals you've invested the effort—and you're legally responsible for everything, not Claude.

I'm a product designer with 20 years of experience across consumer, enterprise, and med-tech products.
Most of my work involves the less glamorous but essential parts of design - information architecture, system design, and making complex products understandable. I've built design systems at different scales and worked across the full spectrum from consumer electronics to clinical trial platforms.
I care about solving the actual problem, not just making things look good. That means understanding business constraints, user needs, and technical realities. I work through flows, architecture, and frameworks to address what needs solving. I'm direct, collaborative, and value teams that make decisions based on insight rather than opinion.
Working across complex products, fragmented workflows and operational ambiguity for nearly two decades
Explore detailed case studies of my projects, showcasing the design process, challenges faced, and solutions implemented.
Launched products and public experiments available to try.
Layer-based pixel editor with cloud storage, cursor and icon exporting
Pixel editor with layer management, cursor and icon exporting, as well as cloud project storage and retrieval.
Import live UI into Figma and export starter code
Browser extension and Figma plugin for importing live UI into Figma and exporting starter code.
See other projects I am working on
Here is what I can help you and your business with
Some of the companies I had the pleasure of collaborating with during my career.
Discover insights, tips, and stories I am sharing from time to time.
AI speeds up building, but every product needs marketing, design, compliance, and care. Quality UI signals you've invested the effort—and you're legally responsible for everything, not Claude.
I spent considerable time working with Claude, one of the most annoying models. It constantly eats up tokens with incessant summaries and generates useless documents. However, past 25,000 tokens and around 500 lines of code, context frays at the edges. All models suffer greatly here—they cannot handle context past a certain point. GitHub Copilot remains my favorite because it's fast at auto-correcting code. That said, models are terrible at remembering what they did.
A short story about how I ended up accidentally building a product that doesn't use AI, while everyone rushes to stuff their socks full of LLMs, generators, and ChatGPT and Claude wrappers.
Have a project, a job opportunity, or just want to say hello?
I typically respond within 24-48 hours with a email and we can set up a call at our mutual convenience.