100% Remote - Computer Vision Engineer - Media Pipe - Human Movement - Mobile App

Senior Computer Vision Engineer — Human Pose & Movement (Mobile)

Full-time contract · 100% remote

The End-Client

A fast-growing consumer health and fitness company, currently transitioning from a content and coaching business into a technology one. They've built a large, highly engaged customer base and a substantial owned audience, and are now investing seriously in product — with an AI-led movement assessment feature sitting at the centre of the app experience.

You'd be the technical owner of the computer vision layer, working directly with the product team.

About the Role

The product analyses a user's movement through their phone camera and returns a structured assessment — joint angles, posture diagnostics, and a weighted score with specific focus areas.

Your job is to make that genuinely accurate and reliable in the real world. You'll own the pipeline end to end: camera and video feed processing, high-fidelity pose detection and landmark extraction, converting those landmarks into precise joint angle measurements (hips, knees, ankles, torso), and building the scoring logic that detects real movement faults rather than noise.

This is full-time, remote, and long-term. You're not consulting on someone else's build — you're the person responsible for it.

What You'll Own

• Reliable on-device skeleton tracking that survives camera perspective shifts, partial occlusion, poor lighting and shaky handheld footage
• Translating landmarks into clinical-grade measurements — hip flexion, ankle dorsiflexion, torso angle, left/right symmetry
• Landmark stabilisation and smoothing so output doesn't jitter between frames
• The weighted scoring engine that turns measurements into a score plus diagnostic feedback
• A clean, documented, extensible codebase the in-house product team can integrate and scale
• The accuracy bar over time — validating output against manual measurement and closing the gap

What We're Looking For

This is a specialist hire. One thing matters more than everything else:

• Deep, hands-on MediaPipe experience applied to human movement, shipped inside a production mobile app. MediaPipe Tasks / Pose Landmarker on iOS and/or Android, running on-device, in something real users have used. This is the core requirement and the first thing we'll ask about.
• Proven on-device inference work — you understand frame rate, thermal throttling, model size and latency budgets on a phone, not just on a laptop
• Strong 2D/3D geometry, camera calibration and landmark stabilisation fundamentals
• You've built pose estimation systems before in sports tech, fitness, biomechanics or rehab — you know the difference between dots on a screen and a measurement someone can trust
• Comfortable owning a technical area outright, making architectural calls and being accountable for the result

Nice to Have

• Python for validation tooling, or model fine-tuning experience
• Background in biomechanics, physiotherapy or strength & conditioning — or a genuine interest in it

Commitment & Contract

• Full-time, 40 hours per week
• 12-month minimum contract with a long-term view — this is an ongoing engagement, not a one-off build
• 100% remote, work from anywhere
• Contractor / B2B / freelance solutions available depending on your country