Frontend Software Engineer – Design System

We're looking for a frontend engineer to help build and grow the Admiral Design System (ADS) — the components, tokens and documentation that teams across Admiral use to build consistent, accessible interfaces.

We care most about the fundamentals, done really well. Accessible semantic HTML and modern CSS do most of the heavy lifting in our components; we reach for JavaScript only when the platform can't give us the behaviour for free. We build small, composable components that extend HTML rather than replace it — not large configurable templates that bury everything in the shadow DOM.

We'd rather expose what the browser already does well than reinvent it, and favour composition (and slots) over sprawling prop sets. We're as careful about what we choose not to build as what we do.

Because our work is used everywhere, the decisions you make travel a long way. A component is built once and consumed across native web, React and Angular — and increasingly by AI tools that build with our system too — so clear thinking about architecture, APIs and long-term maintainability matters more than any one framework. Accessibility and performance aren't things we bolt on at the end;

they're part of what we mean by "finished."

You'll sit at the front of frontend, close to design and UX, helping shape how UI gets built across Admiral and handing components and docs to the product teams who rely on them. We're after someone who's strong on the fundamentals and has good instincts for component design — the years on paper matter less than how you think.

• Building and maintaining ADS components with Stencil.js, documented in Storybook, and distributed for native web, React and Angular.
• Leaning on the platform — semantic HTML, native elements, modern CSS — and adding complexity only when there's a genuine need.
• Making architecture and API decisions that keep components simple, composable and easy to adopt, with downstream impact (versioning, performance, adoption) in mind.
• Partnering with the ADS design team in Figma — shaping component interfaces and helping decide what should be a component, a recipe, or a throwaway example.
• Writing clear documentation for component APIs, usage and accessibility — written once and distributed everywhere, so we never maintain two copies of the same thing.
• Building accessibility and performance in from the start, and proving it (a11y checks, Core Web Vitals, component testing).
• Helping the design system become something AI tools can consume and build with correctly (see below).
• Helping product teams adopt components and unblocking the issues that slow them down.

• Accessible, semantic HTML and modern CSS. This is the foundation, and it matters more than any specific framework or tool.
• Building UI components with clear APIs, sensible composition (slots over large prop sets) and an instinct for keeping things simple.
• Web Components / custom elements
  — understanding how they extend the platform, and where shadow DOM helps versus where it gets in the way.
• Sound judgement on scalable, maintainable component architecture and the downstream consequences of API decisions.
• Working with designers in Figma
  — reading design intent and shaping component interfaces together, including when something shouldn't be a component at all.
• Accessibility to WCAG 2.2 AA
  , and a real feel for performance— bundle size, runtime, Core Web Vitals.
• Stencil.js (or comparable component tooling) and Storybook for building and documenting components in isolation.
• Clear technical writing for a developer audience.

• Consuming web components from React and Angular. (Stencil generates most of the framework wrappers for you, so this is about understanding the model, not hand‑rolling integration.)
• Build, release and distribution: npm packaging, versioning, CI for UI validation.
• Performance techniques: lazy loading, hydration strategies, SSR/pre‑rendering concepts.
• Experience making documentation or systems AI‑consumable — structured metadata, LLM‑friendly docs, agent/MCP workflows.
• Monorepos (Nx, Turborepo) for co‑locating components and wrappers.
• Our wider toolchain:
  Git Hub, Docker, JFrog…