WebAR Game Developer (Three.js + MindAR.js)

Overview

We’re building a lightweight WebAR racing mini-game that runs in mobile Safari (landscape). Players will control a 3D open-wheel car driving around a “floor” play area (implemented via marker-based AR anchoring using MindAR image tracking or similar). MindAR is a web AR library focused on image and face tracking with three.js support, built in JavaScript and leveraging GPU/WebWorkers for performance. - also open to other AR package implementations.

Full time
1 month contract (extendable)

Key responsibilities

• Build a WebAR experience using MindAR (image tracking) anchored to a printed marker/target, with a stable “floor plane” play area.

• Implement car controls for touch (landscape) - steering + throttle/brake (virtual joystick or on-screen controls).

• Integrate and optimise 3D car assets (GLB/glTF), including scale, materials, textures, and draw-call/triangle budgets for iPhone performance. glTF is designed for efficient transmission/loading and reduced runtime processing.

• Implement basic gameplay loop: track bounds, checkpoints/laps/timer, reset, simple collision constraints (MVP level).

• Ensure compatibility and performance on iOS Safari (camera permissions, responsive layout, orientation handling, memory constraints).

• Package and deploy (static hosting, HTTPS), plus basic QA and bug-fixing across target devices.

Required experience

• Strong JavaScript (TypeScript preferred) and modern frontend build tooling (Vite/Webpack, npm).

• Solid three.js experience - scene setup, render loop, cameras, animation, input handling.

• Proven experience with WebAR on the web, ideally MindAR (three.js integration) and the realities of marker-based tracking.

• Experience working with glTF/GLB pipelines and mobile optimisation (mesh/texture optimisation, LOD thinking, performance profiling).

• Comfortable building touch-first controls (virtual joystick, gesture input) and smoothing motion for a game feel.

• Practical understanding of platform constraints - notably that WebXR Device API is not available on iOS Safari, so “markerless hit-test floor placement” is not a given in-browser.

Nice to have

• Experience with MindAR internals (TensorFlow.js-based CV stack) and performance tuning.

• Physics libraries (cannon-es / ammo.js) for more realistic handling.

• Experience with other WebAR stacks (AR.js, 8th Wall, custom pipelines) and knowing when to recommend them.

• Basic 3D tooling skills (Blender) for quick asset fixes and export settings.

• Experience shipping camera-heavy web apps with analytics/error reporting.

Tech stack (expected)

• MindAR.js (image tracking) + three.js + WebGL

• glTF/GLB assets

• TypeScript (preferred), Vite/Webpack, npm

• Optional: joystick/control helper lib

What success looks like (MVP)

• Opens quickly, requests camera permission cleanly, runs smoothly on iOS Safari in landscape

• Stable marker tracking and believable “floor” movement

• Responsive controls and a simple race loop (timer + laps/checkpoints)