Senior UX Designer, Systems Thinker at heart.

Built by developers. Hard for everyone else.

A heuristic evaluation and 8 contextual interviews surfaced three clear patterns: information hierarchy was inverted (business users hunted 10+ minutes daily for project status buried under API metrics), onboarding was undocumented (new hires learned via scattered Loom recordings, 3 weeks to productivity), and there was zero design consistency across the three tools.

System before screen

The hardest sell: spending the first 6 weeks not designing any product screens. Instead — a component library, token definitions, spacing system, and typography scale, with engineering sign-off before a single redesign began.

What this taught me: the first designer on any team isn't there to make things pretty. They're there to establish what good looks like — process, system, quality bar — and make it repeatable.

Built for one side. It had two.

Existing ride-hailing apps optimised exclusively for commuters. Drivers were afterthoughts — 70-80% of the fare after platform cuts, onboarding via 3-day offline paperwork, zero earnings visibility. Commuters, meanwhile, faced unexplained surge pricing and inaccurate tracking. 12 contextual interviews (6 drivers, 6 commuters) plus ride-along shadowing shaped the direction.

Show earnings, not fare

Research showed drivers think in "what I keep," not "what the passenger pays." A co-design session with 4 drivers confirmed it — they universally sketched fare and route first, never the passenger name.

Two-sided platforms are two different products. The design system unifies driver and commuter apps; the UX has to serve each on its own terms.

Three products, three teams, zero shared language

Each vertical grew independently with its own conventions. A fan moving from watching a live fight to reading a fighter's comic to buying merch felt like they'd switched apps entirely. An audit found 200+ unique components, 14 conflicting colour definitions, 6 spacing scales.

Fans think in fighters, not verticals

40+ fan sessions across Southeast Asia revealed the product was organised around internal company structure, not fan behaviour. The fix — a fighter profile page connecting live fights, comics, and merch — wasn't in the original scope.

A shared token system, built before any screen design, saved an estimated 6 weeks of rework — and outlasted the engagement, still maintained by the ONE team today.

Designing for agents, not just for humans

Warble Cloud is where I'm actively developing the AI/agentic UX muscle: interfaces have to represent an agent's reasoning process, surface confidence and uncertainty, and give an on-call engineer enough context to trust — or override — an autonomous decision in production.

In parallel, I'm learning Python, Electron.js, and Kubernetes fundamentals — not to become an engineer, but so I understand what an agent can actually reason about, fail at, or hand off, and can design accordingly.

This project is live and evolving — a fuller case study with research, decisions, and outcomes will follow as the platform matures. Reach out if agentic UX is something you're hiring for.