Creator–brand collaboration is full of friction: discovery, briefs, negotiation, delivery. The temptation is to build for all of it at once — and end up with a platform that does ten things adequately and nothing memorably.

I worked with the founders and engineering to narrow the surface area: define the wedge, sequence the roadmap, and redesign the core collaboration flow so the AI does real work — matching, drafting, pricing — instead of decorating the UI.

Day to day that means owning specs, running the design, and keeping the team honest about what's actually for the user versus what's for the deck.

Live and iterating. The product has a clearer story, a tighter onboarding, and a roadmap the team can defend.

I led product across everything Chineasy shipped — the app, a full website relaunch, even a board game — from planning and scope through testing, deployment, performance, and the day-to-day operations after launch.

The app was a 2022 Apple Design Awards finalist, reached No. 2 in the UK App Store (editor's choice) and No. 6 in the US education category. Beyond the app: a podcast featured by iTunes and Amazon Alexa, and a board game that won a Parents' Choice Award.

Companies want to support employees becoming parents but reach for generic gifts. The opportunity: a brand built around the real moments of becoming a working parent — and finding your feet again at work — that makes people feel seen.

Positioning and messaging, product curation, the brand voice, cold-outreach playbooks for US brands, and the unglamorous operations: orders, fulfillment, and the handoff to logistics partners.

Being 'Employee No.1' means there's no one to delegate the boring parts to — so the strategy has to survive contact with reality. It keeps me honest.

Growing its US client pipeline, with a repeatable outreach motion and a brand that holds together from first email to delivered gift.

I run several projects as an operator and a parent — time is the scarcest thing I have. Existing tools optimize for teams and tickets, not for one person trying to protect deep work. So I built the tool I needed.

Building it solo — design, SwiftUI, the data model, the visual system — is the most honest product education there is. Every shortcut I'd ever waved through as a PM, I now had to actually live with.

It's also where I practice building in public: shipping, writing about the decisions, and letting the work be the proof.

In active development — a Partner → Project → Task model, Workspace Focus, and a soft, rounded visual language that makes daily use feel calm rather than nagging.

The whole thing, end to end: bilingual landing pages, the application and registration flow, payment, a tuition and scholarship calculator, and the backend pipeline — Cloudflare Workers → Notion → email — that runs admissions automatically.

Most strategists hand the build off. I don't. When a program needs a working system, I ship it myself — which means my product decisions are never just theory.

A large-scale youth summit under Dots Global. I led the program coordination — aligning speakers, partners, and the team to put it on, on schedule.

A government-funded capacity-building program — internal operations and project management across multiple stakeholders, keeping deliverables and deadlines on track.

This is the counterweight to the product work — proof that strategy without operations is just a nice document. New programs land here as they launch.

Three surfaces on one deploy: a public page for admissions and logistics; a participant area where students sign in with a personal code to check the schedule, request leave, book a make-up session, and fill in the pre- and post-class surveys; and a staff console behind a passcode for attendance, announcements, rescheduling, and collecting submissions.

A Cloudflare Worker sits in front of Notion as the datastore — roster, leave and make-ups, pre-class surveys, post-class feedback — with R2 holding the documents students upload to justify an absence. Keeping the records in Notion means the program team can read and correct them in a tool they already use, instead of waiting on me.

Same pattern as the camp registration system: the program needed working software, so I shipped it rather than writing a spec for someone else. Running the program and building its tools at the same time means the tools match how the work actually happens — not how it looks on a plan.