Front-End Engineer Design System (medior/senior)

A small, autonomous team inside Northwind Robotics, building the design system and global header that every other engineer and customer touches. Overview: Workstyle: This is a hybrid position; however, to be eligible for hire, candidates must reside within the Czech Republic, with in-office presence for onboarding and on agreed days afterwards. Expect roughly one hour a day of synchronous time - huddles, quick syncs, occasional pairing. We ask for camera on and full focus during that hour. Some things are genuinely faster in sync. What we're looking for: Engineering judgment that goes beyond the code. You think about how your components will be used, how they'll evolve, and what happens when someone uses them incorrectly. React and TypeScript are the baseline. What matters more is how you decide what to build and why. Real fluency with agentic coding tools. Not "I've tried Copilot." Actual experience using AI to ship, and opinions about where it helps and where it doesn't. Design system instinct. You notice when a prop won't scale, when a presentational component is doing layout work, it shouldn't, and when an API will frustrate the next engineer who touches it. Background in Computer Science or Software Engineering preferred. Self-taught is welcome if you can show the work - a personal project, a contribution history, something concrete. English B2 or higher. Minimum of 5 years of experience. Participate in an on-call support rotation, providing prompt response to production incidents. How we work with AI: You use Claude Code (or equivalent) every day. It writes the code. Your job is the layer above it: understand what needs to ship, direct the agent to build it, verify the output is actually correct, and take responsibility when it isn't. We treat the agentic development lifecycle as an evolving practice; spotting and shipping improvements across discovery, testing, deployment, and QA is part of the role. "Correct" is not "the tests pass." Jest, Storybook, visual regression, token usage - agents check all of that, and they do it well. We need a human who holds the line one level above: does this actually solve the problem, does it fit the system we're building, will it harm internal or external users. The discipline we're looking for is the discipline of reviewing outcomes - what's actually shipped, how it behaves, whether it's the right thing at all - rather than the discipline of reading every line. What you'll do: Build the DX tools that other engineers at Northwind Robotics rely on every day. Design System components with React, TypeScript, Storybook, Chromatic, Playwright. Clean component APIs, a 3-tier token system, and an MCP server that keeps documentation in sync with the code. Own our global header/footer microfrontend. Technically interesting: SSR with multi-layer caching, feature flag orchestration, universal deploy across multiple consuming apps with different rendering strategies. The first thing every customer sees is a real architectural puzzle. Push the agent development lifecycle forward. We already have our own MCP tooling and an LLM-friendly robots file; release automation is next. Work independently. You have Claude, the internet, a well-documented repo, and our internal knowledge base. Investigate first, then come up with a proposal, not a question. On process and business questions, raise them early. How you'll make a difference: Northwind Robotics helps people celebrate moments together - birthdays, team events, family gatherings. The global header is the first thing every customer sees and the primary navigation they use. Getting it right is one of the most direct customer impacts you can have on the site. Our Design System is the foundation that other engineers at Northwind Robotics build on. Clean component APIs, accurate migration notes, and an MCP server that documents itself - what you ship makes every product engineer's daily life materially better. Internal tools are products; the people using them are your customers. How you'll be measured: Quality of what ships from your desk. Not lines of code or velocity - whether what you ship solves the actual problem, fits the system, and doesn't harm internal or external users. How well you navigate ambiguity. Whether your default move is to investigate and come back with a proposal, not a question. How you use AI as leverage. Whether you direct agents well, verify their output critically, or implement your own ideas on how to improve the shared AI prompts and workflows. Who this role is not for: Someone who prefers clear instructions over open-ended ownership. We trust people to make their own calls, and the role requires you to set direction, not wait for it. Someone who treats AI-generated output as finished work. The judgment on top is the actual work. Someone who needs ambiguity resolved before they can move. You investigate, decide, and adjust as you learn. Someone who shows up passively in sync time. Our hours together are short, and we use them actively. C