A new way to
buy insurance.

Traditional insurance only came one way — a year-long policy, bought through a broker, locked in for twelve months. But independent workers were hiring out by the gig, by the day, sometimes by the hour. Thimble built a platform where they could buy short-term, flexible coverage on demand. Insurance that matched how work actually happened.

A foundation built
once, used everywhere.

As founding designer on a team of seven, I owned the full design framework — the transactional mobile app, the marketing site, the visual identity, even the deck that closed our Series A. Working directly with the CEO and early stakeholders, I helped translate a new SaaS model for episodic liability coverage into something users could actually grasp. Every surface had to feel like the same product. I built the early system around three principles — plain language over jargon, momentum over completeness, and one decision per screen — and applied them consistently across mobile, web, and email.

Stream and subscribe
your policy.

To become a one-stop shop, we kept expanding — more jobs covered, more product lines, and a new subscription plan. Users could pay monthly, modify their coverage, pause when work slows down, and cancel anytime. Insurance you could use the same way you use a streaming service.

From a blank canvas
to a category disruptor.

A new channel,
a new kind of user.

Brokers started using Thimble on their own, placing coverage faster than traditional carriers could. Together with PMs, customer support, insurance, and BD, I ran broker interviews and audited their workflows: dozens of clients in motion, multi-carrier spreadsheets, fifteen-minute calls that needed a quote on screen by minute three. The opportunity was clear. A D2B2C channel could expand revenue without diluting the consumer product. But brokers needed a product built for how they actually work.

See what's possible,
then sell it.

The hardest part of selling insurance isn't writing the policy. It's figuring out what to write. The Appetite Checker takes a few inputs about the client's business and surfaces every coverage Thimble can write, including GL, PL, Cyber, IM, and WC, along with the flexible durations Thimble is known for. From there, brokers move straight into quote and bind. No dead ends, no second guessing, no leaving the page.

Quote, bind, manage
in one place.

Beyond a single sale, brokers run an ongoing book of business. The portal brings every client and policy into one workspace where brokers track activity, manage renewals, handle endorsements, and scale from a solo agent to an agency of 200 with role-based permissions and agency-level reporting.

A closet full of clothes
and nothing to wear.

Jay and I weren't trying to build a company. We were trying to solve our own problem. Most people own dozens of clothes they barely think about. Every morning, they pick from the same five things on top. Wardrobe apps existed, but they all looked like spreadsheets with photos. We wanted something that felt like a tool you'd actually open. No funding, no investors, no roadmap longer than a weekend. The whole product came down to three moves: add the clothes you love, plan outfits, build your wardrobe.

Where design meets engineering,
something has to give.

A year of working with one engineer taught me lessons no team could have. We disagreed often — sometimes for weeks — about whether a feature was worth building. The fix wasn't picking a winner. It was sitting down, talking through trade-offs, and finding a third option together. The harder lesson was time. Without a client or deadline, the project stretched longer than it should have. Even a passion project needs milestones. "We'll get to it" is the most expensive sentence two collaborators can say.

A side project
that kept growing.

Nostalgic. Retro.
American.

The 1950s and 60s supermarket era is a rich visual language — bold letterforms across serifs, sans, and scripts; hand-painted signage; saturated produce posters. I built the identity around a small set of retro signals: a confident mix of weights and styles, a strong color pairing, and decorative supporting marks, held together by quiet modern typography and generous whitespace.

A quiet job,
done.

Good retail design disappears. The store stops feeling designed and starts feeling found. That was the bar I set: not loud, just inevitable on the block. A corner of Midtown that finally felt like it had always been there.

One dome,
then many domes.

Everything Studio's identity is built around the geodesic dome — Buckminster Fuller's icon of structure achieved through diversity of parts. The studio's whole practice runs the same way: many disciplines, one shape. The invitation needed to say something about how Everything Studio thinks before anyone had clicked into the event details, so I started where the studio starts — its own logo — and let it multiply. A single dome became a field of them, each one a slightly different angle, weight, and rhythm. The visual is the lecture in miniature: one studio, many ways of working, held together by a recognizable form.

A small project
on a large stage.

It was a short brief, an early credit, and a room I'm still proud to have designed for. The lesson stuck: when a studio's identity is strong enough, you don't need to invent a new visual language to speak with it — you just need to know how to make it talk.

A home screen
with personality.

Phone home screens are static by default — rows of small rectangles that exist to be tapped, not to be looked at. Transfoner pushes against that. Each icon becomes a small character that breaks apart, transforms into objects, and reassembles. The phone surface stops being a grid of tools and starts behaving like a living scene.

njenworks → Pentagram.

Transfoner now lives in Pentagram's archive.

Masks made to scare,
and to protect.

Like many other cultures, Korea has its share of superstitions and old beliefs — woven into villages, regions, the rhythms of the year. Ancient Koreans carved wooden masks of local spirit figures and wore them in ritual ceremonies and theatrical performances. The faces were extremely bizarre and grotesque on purpose — meant to ward off evil spirits, and to bring good fortune to families and neighbors. Drunken Rice Bomb borrows that same logic, pairing six rice-harvesting regions with six of their traditional folk spirits, one mask for each.

A little good
fortune, with a buzz.

The masks did their job for centuries. Drunken Rice Bomb hopes to carry a little of that forward — warding off the bad and welcoming the good, with a little buzz along the way.

Always more to see
— let's talk.

Building the bridge.

As the team's first designer, my main focus for the holiday launch was the parent app, end-to-end in four months. Onboarding, setup, dashboard, chat monitoring, settings, family sharing. The design foundation grew alongside — a system to tie every future feature together, so as bondu grew, it would still feel like one product.

The entire bondu experience is COPPA-compliant, the US child-safety standard. Every play conversation runs through safety filters. And through the app, parents can monitor chat, get summaries and alerts, set guardrails, and delete anything anytime. Safety, accessibility, and usability — the three principles behind every design decision.

A business that can
carry the cost of AI.

AI doesn't run for free. LLM costs are real, and runway is tight. Just weeks after launch, we shipped a subscription model: Free vs Plus, at $19.99 a month. Plus opened up more talk time, in 27 languages, and the full library of play features — Personas, Bedtime Mode, Immersive Stories, Motion-responsive play, and more.

"Scale" would imply natural growth. This was a deliberate bet to turn a one-time toy purchase into a recurring product, while keeping the toy itself accessible.

From product to brand
to funnel.

As the product matured, the design work moved outward. I led the redesign of bondu.com, the parent-facing site that doubled web conversion. Social posts, ad creative, brand voice for the parent audience. Same brand, new surfaces.

Four phases,
four numbers.