I'm a product designer who ships her own code — mornings for sketches and prototypes, afternoons for the React that makes them real.
For the past eight years I've helped studios and early-stage teams take rough ideas from the first whiteboard to something people pay for.
Hi — I'm Mara Voss, a product designer and front-end developer working from a small studio full of paper, sticky notes, and exactly one healthy plant.
I fell for this work the first time a sketch of mine became a screen someone else could tap. Ever since, I've chased that moment: the point where a drawing stops being a picture of software and starts being software. Designing the interface and writing its code are, to me, the same craft held in two hands.
Code is only half the notebook, though. I keep a long-running set of studio notes — short essays on process, CSS, and the strange politics of buttons — because writing a thing down is the fastest way I know to find out whether I actually understand it.
If there's a thread through my projects, it's restraint. I like quiet typography, honest defaults, and interfaces that don't perform. A screen should feel like a well-set table: everything within reach, nothing demanding applause.
This site is where the three strands meet — the projects I've shipped, the notes I've written, and a way to reach me if you'd like to make something together.
Thanks for stopping by. The kettle's on.
A calm money app that shows freelancers what a slow month actually costs — before it arrives.
Freelancers don't fail at budgeting because they're careless; they fail because every tool assumes a salary. Ledgerline starts from the opposite premise: income is lumpy, and the interface should absorb the lumps instead of scolding the user for them.
I designed and built the product end to end — research calls with eleven freelancers, the interface system in Figma, and the production front end in React. The centrepiece is the runway view: one quiet line that answers "how long can I keep saying no to bad clients?" without a single red warning badge.
The hardest design decision was what to leave out: no streaks, no confetti, no daily nudges. The app earns trust by staying quiet — retention came from usefulness, not notifications.
A field-notes journal that turns a messy camera roll and a few scribbles into a small printed travel book.
Every trip ends the same way: four hundred photos, three voice memos, and a note that just says "that bakery??". Fernpost is built for the week after you get home, when the memory is still warm but the energy to make an album is gone.
The app clusters photos by place and day, asks one gentle question per cluster, and lays the answers out as spreads. I wrote the layout engine in TypeScript — a constraint solver that keeps captions with their photos the way a careful person would, not the way a grid template does.
My favourite detail: the app never says "empty". A day without photos becomes a quiet page with the route line and a space to write, because those days are part of the trip too.
A pocket-sized component kit that lets a team of two ship interfaces that look like a team of twenty.
Most design systems are written for companies with a design-systems team. Gridfolk is written for the other 99% — two people, one repo, no time for governance meetings. It fits in your head: twelve components, one spacing scale, and a single CSS file you can read over coffee.
I maintain it as an open kit and use it as the bones of my client work. Every component ships with its reasoning written down — not just how the button looks, but why it refuses to be smaller than a fingertip.
The kit's rule of thumb is stolen from cooking: mise en place. Decide the tokens before you're hungry, and every screen after that is assembly instead of invention.
Short essays on design, CSS, and code — written to find out whether I actually understand what I think I understand.
May 14, 2026
Every few months someone rediscovers that you can skip mockups and design straight in code, and every few months a project quietly drowns in half-decided details because of it. Both things are true: the browser is the best design tool we have, and it is also a machine for postponing decisions.
The failure mode is subtle. In a drawing tool, an unresolved question looks unresolved — the spacing is obviously fake, the copy is obviously placeholder. In the browser, everything renders. Unfinished thinking gets a border-radius and starts to look like a choice.
What keeps me honest is embarrassingly small: before I open the editor, I write five sentences on paper. What is this screen for, who arrives here, what must they read first, what may they ignore, and what happens when it goes wrong. Five sentences, then code.
With the sentences pinned above my desk, browser-first design becomes what it promised to be — real text, real widths, real states, and no translation loss between the picture of the thing and the thing.
/* Decide these before you're hungry. */ :root { --space: 4px; /* one rhythm, multiplied */ --ink: #404040; --muted: #737373; --line: #e5e5e5; }
The tokens file is the paper sketch's twin. If a value isn't in it, I'm not allowed to type it. That single rule converts "designing in the browser" from improvisation into composition.
March 2, 2026
Big-company design systems solve a coordination problem: forty people must not invent forty buttons. A team of one has no coordination problem. What it has is a memory problem — you at 11pm cannot remember what you at 9am decided, so you decide again, slightly differently, forever.
So a solo system isn't a rulebook; it's a shelf. A place where past decisions sit within reach, labelled well enough that reusing one is easier than remaking it.
Everything else from the enterprise playbook — contribution models, versioning policy, review boards — is weight without lift at this scale. Skip it guiltlessly.
The test of a one-person system is simple: when a new screen takes an hour instead of an afternoon, and looks like it belongs to the same product, the shelf is working.
January 20, 2026
Here is a small experiment. Design a confirmation dialog with the button label "OK", then redesign it with the label "Delete 400 photos". The second dialog wants different spacing, a different emphasis, probably a different color. The words changed the layout — which means the words were doing layout work all along.
That's why I write interface copy before I draw the interface. Not because the words are precious, but because they are structural. The label is the load-bearing wall; the padding is drywall.
None of this requires being a great writer. It requires only the admission that the interface is mostly made of sentences, and that sentences, like grids, can be designed.
Have a project in mind, a question about a note, or a bakery recommendation? Write to mara@voss.example or use the form — I read everything.
Eight years of designing and building products — first as the designer in the room, lately as the person who also merges the pull request.
2023 — Present
Principal Product Designer
Design partner for early-stage founders
Led design for six product launches, from first research call to shipped v1
Built the studio's shared component kit, cutting prototype time roughly in half
Paired weekly with client engineers so design intent survived the handoff
Mentored two junior designers into confident front-of-room facilitators
2020 — 2023
Senior Product Designer
Consumer fintech, from seed to Series B
Redesigned onboarding, lifting activation from roughly half of signups to four in five
Owned the design language through two product pivots without a full rebrand
Introduced usability sessions as a weekly habit rather than a quarterly event
Shipped the accessibility pass that took the app to WCAG AA
2018 — 2020
Product Designer & Front-end Developer
Small agency, big variety
Delivered fourteen client sites end to end — design, build, and launch
Moved the agency's starter stack to accessible, componentized templates
Ran client workshops that turned vague briefs into testable screens in a day
2016 — 2018
Visual Designer
Where the typography habit started
Designed identities and editorial layouts for twenty-odd small brands
Taught myself HTML and CSS to stop shipping designs I couldn't defend
Built the print-to-web checklist the studio still uses today