Experience · 15+ years
LI.FI & Jumper.xyz · Director of Design
- One designer served a company scaling from 16 people to over 130. Hiring alone was never going to close that. Grew the function to a 7-designer org reporting to the CEO, and championed AI adoption beyond design into product and business development, with Claude Code, Codex and Figma AI.
- Design feedback was ad hoc, scattered across Slack, Telegram, Discord and DMs. Built two standing Workshops a week, designers critiquing each other on Tuesdays and presenting to product and engineering on Thursdays, plus weekly one-on-ones.
- An outside agency left a token-swap widget that did not work. Scrapped it, rebuilt from scratch within months, then grew it past swapping into portfolio tracking and a gamified XP system: it spun up as its own brand, Jumper.xyz, and platform volume went from $0 to $3B+ monthly.
- Partners had to integrate the Widget before they could see it in their brand. Shipped Playground to show it live and Theme Composer to skin it - the demo became the B2B team's sales tool.
- LI.FI had outgrown the brand it launched with. Redesigned its mark in 2025 - still in use company-wide - built the 2.0 system around it, and created Jumper's identity from scratch, so one company could serve B2B and B2C.
- Design, engineering, BD, marketing and ops each worked from their own references. Directed a public website that doubles as the internal design-systems site, documented so any team's AI could read it. BD produced on-brand proposals in hours instead of days, without a designer in the loop.
PayPal · Sr. UX Designer, Revolving Credit
- Inside the new PayPal App, screens were drawn file by file, so one change had to be re-applied screen by screen across multiple products. Created the Assembly Design Framework on PayPal UI, where a screen is composed rather than drawn: a single published edit now propagates to every mockup in every file. Teams published their newest components there for everyone to use, and many reached PayPal's official UI Library through Assembly rather than the reverse.
- Prototypes were not realistic enough to test on. Added persona-driven module stacks and a readiness checklist, so the same system turned out research-ready prototypes on one source of truth shared by product, engineering and content.
