Staff Design Engineer

tl;dr: We are looking for a Staff Design Engineer who thinks like a designer, codes like a front-end engineer, and treats AI as a collaborator.

Scribe is where exceptional people come to do the best work of their careers. More than 94% of the Fortune 500 use Scribe to own their specialized intelligence: the unique way their teams work, decide, and get things done. Our Specialized Intelligence platform automatically captures how work happens and turns it into a living asset that helps people and AI agents do their best work.

We're growing fast. Since our founding in 2019, we've grown to 7 million users across 600,000 businesses. Based in San Francisco, we've been named a LinkedIn Top Startup, are valued at over $1 billion, and are backed by leading investors. Join us in our mission to transform how people do work.

We're builders who hold a high bar, move fast, and care deeply about each other and our customers.

We're hiring a Staff Design Engineer to own Stylus, our design system, and the quality of what we ship across the web app, desktop app, browser extension, and marketing site. As a Staff Design Engineer at Scribe, you sit at the intersection of design and frontend engineering.

AI has made designs, code, and prototypes cheap to produce. What separates good product from great product now is judgment: knowing which interactions are worth building, and then building them well. We want someone who can look at a problem, decide what it should feel like, and ship it themselves.

At its core, this role is anchored in Stylus ownership. From there you'll drive both design craft (product polish, prototyping, and interaction quality) and platform power (frontend tooling, DevEx, and the internal agents and skills).

• →

• Land your first Stylus change in the first week or two - small, real, in production. You'll learn more about our stack from one merged component fix than from a month of onboarding docs.

• Build the honest picture: where components have drifted, which surfaces are furthest from the system, and what our accessibility coverage actually is versus what we assume it is.

• Meet the people who file the requests - designers, product engineers, marketing - and watch customers use the product. Craft decisions get better when the person making them has watched someone struggle.

• Figma plugins, automations, or CI checks that report design system adoption and accessibility

• Web components or multi-framework experience, including shipping a system into non-React surfaces

• Desktop apps or browser extensions, especially always-on ones

• Agent-facing or generative UI work

• Publish the Stylus roadmap and the quality bar that goes with it, and start using both to say no. This is the point where a backlog becomes a plan with an owner.

• Pick the first hard component and build it - the one where aesthetics and performance fight each other, that nobody has had the time to do properly.

• Start the legacy migration where it pays off most, and make the path off legacy obvious enough that other engineers take it without being asked.

• Get adoption, accessibility, and documentation health reported automatically, so system health is a number we see rather than a question we ask.

• Make contributing to Stylus something engineers and designers actually do: a clear way in for requests, contributions, and exceptions, with you making the call on each.

• Ship the first AI workflows you'd want the rest of us using - the prompts, skills, and tooling that let designers and PMs build against the system safely.

By six months: Stylus has a real roadmap instead of a backlog, the legacy-component migration is actively moving, and adoption is reported automatically rather than asked about.

• 8+ years of experience in front-end development with a strong design sensibility and real depth in CSS, the DOM, and browser rendering. You have a view on when to use native platform elements versus a framework's version, and you can explain it.

• Design ability of your own. You can make the call on an interaction, and you know which ones aren't worth what they cost to build.

• You've worked on a design system at depth: components, primitives, tokens, versioning, adoption, and the governance to hold a line when a team wants an exception.

• You shape the work rather than take tickets. You'll be accountable for a roadmap, which means knowing when to say no.

• You already build with AI tools like Cursor and Claude Code, and you like getting other people using them well.

• Accessibility, keyboard and touch input, and respecting user preferences are things you handle by default.

If you're reading this thinking "that's me!", we want to meet you.

Our loop is a code-and-craft exercise, not a portfolio review.

Hybrid: Based in San Francisco. We work from the office 3 days per week.