Front-End Engineer (React Three Fiber · AI-Assisted Development)
US client via staffing partner · Hoboken, NJ
- Location
- Hoboken, NJ · Onsite
- Salary band
- Band not stated
- Type
- Contract
- Level
- Senior
- Work authorization
- No sponsorship — now or in future
Stack
React 19 · React Three Fiber · Three.js · Zustand · TypeScript · Vite · WebGL · GLSL · ARIA · WCAG · accessibility · screen-reader testing · GPU/CPU profiling · real-time performance optimization · glTF/GLB · Claude Code · AI-assisted development · keyboard navigation · focus management · state management
About the role
Contract || Front-End Engineer (React Three Fiber · AI-Assisted Development) || Location: Hoboken, NJ - Fully Onsite Note- Only W2 , No C2C About the project We are building a browser-based 3D virtual science lab. Students assemble apparatus, perform experiments, and are assessed on their technique — all rendered in real time with React 19, React Three Fiber (Three.js), and Zustand, built with Vite. The simulation runs on ordinary school hardware, including low-end Chromebooks, so it has to be both convincing and fast, and its behavior has to be deterministic enough to score reliably. What you'll do - Build and own complex React Three Fiber features: interactive 3D equipment, a selection-driven action system, real-time visual effects, and the UI around them. - Keep the app fast on low-end hardware — profile and tune the render loop, manage GPU/CPU cost, and hold a frame budget on Chromebook-class devices. - Keep application state coherent as it grows (simulation state, equipment state, an event log used for assessment, and UI state) using Zustand. - Make accessibility real: full keyboard operation, focus management, ARIA, and screen-reader support are core requirements here, not polish. - Direct Claude Code effectively — decompose work into well-specified, agent-sized tasks, review the output with a senior eye, and know when to write something by hand instead. - Help maintain the conventions, CLAUDE.md guidance, and repo structure that keep AI-generated code consistent and safe. - Collaborate with our 3D technical artist to integrate assets and shaders cleanly. Front-end engineering (the foundation — non-negotiable): - Deep, senior-level React (React 19 / modern React): the render model, hooks discipline, memoization, effect timing, and a real feel for where re-renders and stale closures come from — which matter acutely in an app running a physics loop every frame. - Three.js and React Three Fiber in production. Real command of the scene graph, geometry and materials, the useFrame render loop, and disposal / memory management (WebGL leaks are a real hazard). This is the heart of the role and the skill we weight most heavily — React experience without R3F is not a substitute. - Real-time performance discipline. You think in frame budgets, know what drives GPU vs CPU cost, can profile a render loop, and habitually design for the weakest target device. - State management at scale with Zustand (or clearly transferable experience), keeping a growing state model coherent rather than tangled. - Accessibility as a first-class skill — keyboard navigation, focus management, ARIA, WCAG AA, screen-reader testing. For an education product this is a core requirement and a procurement gate, not an afterthought. - TypeScript, testing, and modern tooling — TypeScript fluency, Vite, a real testing discipline (we favor testing real behavior over mocks), and the judgment to set the quality bar for others. Working effectively with AI coding agents (the differentiator — evaluated seriously): - You can decompose work into agent-sized units — scoping a task with enough context, clear boundaries, and defined done-conditions so an agent produces good work. You write specifications, not wishes. - You review AI output critically — reading every diff with the skepticism you'd bring to a junior's PR, catching the plausible-but-wrong, the convention drift, and the code that passes tests but violates an architectural rule. You never ship output you don't fully understand. - You know when not to use the agent — a feel for which tasks fit an agent well and which are faster or safer done by hand. - You are comfortable with the operating environment around the agent: context management, CLAUDE.md / skills / MCP, and structuring a repo and its conventions so agents produce consistent results. (Multi-worktree / parallel-agent experience is a strong plus.) - You have honest calibration about the tool — neither refusing it nor over-trusting it, with real intuition, from experience, about where it's strong and where it approximates or fabricates. Strongly preferred: - WebGL / GLSL literacy — enough to collaborate with a technical artist and debug shader and asset integration. - glTF / GLB loading and the asset pipeline. - Prior educational, assessment, scientific, or simulation software experience.
Recruiter contact
Just $1 to see this recruiter's name, email and phone. One-time, per role — no profile needed.
Want the full job-search package instead? Talk to us →
The recruiter's details are not public. Helen makes the introduction, checks your resume against the requirements first, and follows up on the reply.