40% → 0.3% detach rate
Figma components used to drift from code in 40% of cases. After the FORK token architecture, detach dropped to 0.3% across 8 products.
Tina Singh / UX + AI practice
UX + Systems Designer. Ships with AI.
Senior UX/UI Designer who builds with AI and owns the quality. Five case studies showing how I use AI for the heavy lifting while keeping design judgment, security, and QA in human hands.
Selected work
Five problems.
Each one backed by a number, an artifact, or a failure on record. The war stories tend to be more useful than the wins.
01+3 tags
Sentinel
A local context harness that manages what AI sees and versions everything in git. Built through conversation.
02+3 tags
Fast A11y Pipeline
A11y backlog was blocking revenue. Built a pipeline with Claude Code so fixes could ship without pulling engineers off their roadmap.
03+3 tags
Self-Review
Asked Claude to evaluate its own work. It omitted the two worst mistakes. Now I audit the audit.
04+3 tags
FORK
Stopped tokens from chaining through each other. 94k in production. 0.3% detach rate.
05+3 tags
This Portfolio
Built this site with AI. Found four bugs Claude introduced and zero it caught. This page is the proof.
By the numbers
The work is measurable.
Every claim has a number or an artifact behind it. If a sentence could appear on anyone's portfolio, it gets cut.
Design decides. AI executes.
When AI breaks the same rule twice, I stop prompting and write a constraint document instead. 36 reusable skills that enforce design intent structurally.
94k+ design tokens live
Design tokens powering 8 products, with 369 Figma files scanned bimonthly for drift. 15 component specs written in agentic format so AI tooling can consume them directly.
Get in touch