A Portfolio
Twelve years of interfaces, set in order of conviction.
Manifesto
I design interfaces the way a typesetter sets a page — every element earns its millimetre. Restraint is not a style; it is a decision made a thousand times an hour. I am suspicious of delight that asks nothing of the user, and I would rather a product be quietly legible than loudly impressive.
Product Designer — Interfaces, Systems, Soft Realism. Helping software teams ship products that read like they were edited, not assembled.
§ 01 — Trusted by
§ 02
Six projects — 2021 / 2026
A calm operating layer for distributed teams — six surfaces, one grammar, zero notifications you didn't ask for.
Outcome — cut average time-to-context from 11 min to 2 min across 14 partner teams; notification fatigue down 63% in the first quarter. Now the reference architecture for the org's internal tools.
A type-driven design system for a literary reading app — scales that respect the eye, not the grid.
Outcome — 240+ tokens, one variable axis; average reading session +22%. Adopted across three sister products.
A field-research tool for environmental scientists — capture that survives mud, signal, and dead batteries.
Outcome — works offline to 72 hrs; data loss in transit fell from 8% to under 0.4%. Used by 11 university field programs.
A publishing platform that hides share counts — the metric was never your friend.
Outcome — median read-depth +41% in six months; 3,200 writers migrated in the first year.
A wayfinding system for a regional transit network — 47 stations reduced to one glance, legible at a run, in the rain, without your glasses.
Outcome — missed-transfer rate down 19%; the system is now the case study the network shows visiting delegations. Rollout completed in 14 months.
An onboarding flow for a fintech that asks for less up front — and converts more by doing so.
Outcome — activation +34%, drop-off at step 3 halved. The four-field pattern became the org's default.
Work shipped with teams at Linear, Plaid, and two studios you have not heard of yet — both of which are the better for it. Currently building quietly; two engagements per quarter.
§ 03
Where I am sharp, and where the work actually lands
End-to-end ownership of a product surface — from the first uncomfortable question to the pixels a user actually touches. I design the thing, not the deck about the thing.
Tokens, primitives, and the grammar between them — built to be authored by many and read by everyone. A system earns its keep when a new designer ships correct work on day two.
Code-level prototypes that answer the real question before the meeting — interaction, motion, edge state. I would rather build it twice in code than argue about it once in a frame.
Generative interviews, usability, and the unglamorous work of reading support tickets. I treat research as the brief, not the decoration on top of one.
§ 04
Four movements, repeated until the product is quiet enough to ship. Not a methodology — a rhythm.
Before any pixel: interviews, support logs, the data nobody reads. The deliverable of this phase is a single sentence the team agrees is the problem. Most projects are solved here, not later.
Cheap options first — paper, lists, ugly frames — generated against the real limits: engineering cost, latency, the legacy nobody wants to touch. I would rather kill an idea in an hour than in a sprint.
A working prototype in code, put in front of five users before the design is "done." The second build is always cheaper, and it is the one that ships. Motion and edge states get designed here, not bolted on.
The last 20% is subtraction. I remove what the prototype proved we didn't need, write the spec the team can own without me, and document the system so the next designer isn't reading my mind.
§ 05
Lin is the rare designer who can hold the whole product in one hand and a single pixel in the other, and not drop either. Halcyon shipped on time because she refused to let ambiguity survive a meeting.
“She redesigned our onboarding and our activation curve bent the same week. I have never seen a designer move that fast without making a mess.”
“The type system she built is the reason three of our products finally look like they came from the same company. It just works.”
§ 06
Awards, talks & press