NNg Certified - ID 1064178
Solo builds that test how I think about AI, systems and trust. Each one is live, and each page says what I decided and where it falls short.
One design system behind four live products, so a change happens once instead of four times.
Audited about 5,936 CSS declarations first, then built 236 tokens and migrated every product onto them. Shared components arrived in v0.2.0, the display font changed in v0.3.0 and a light theme arrived in v0.4.0. Honest limit: only the portfolio uses the components and the theme; the other three products still keep their own component copies.
A one-line field on the home page that answers questions about my work, with a link to the case study behind each answer.
Answers come from a private knowledge file and a short list of rules: exact numbers, no borrowed credit, and a polite no when the file doesn’t say.
A weekly briefing on what UX practitioners are debating, generated from design publications every Monday.
Rebuilt after the first version showed status it didn’t have. Code now rejects any briefing that cites outside its group, and a failed run leaves last week’s in place.
A public demo of the UX auditor agent I built for an enterprise Kiro AI kit: paste one component, get findings against NN/g’s 10 heuristics.
It reports only what it can point to in the code, five findings at most, and says so when nothing qualifies.
Paste a job description and get a 0–100 match score against my background, the gaps, and a compensation signal.
Switched from Sonnet to Haiku to fit a 10-second hosting limit. A decision-support tool, not a predictor.