Senior Staff Software Engineer in Sheffield

Fyous is building a new class of manufacturing technology — fast, digital, zero-waste, and fundamentally different to anything on the market today. Our PolyMorphic Moulding system uses tens of thousands of digitally controlled pins to form a mould in minutes, thermoform a part directly over it, and then reconfigure into the next shape. No resin. No waste. No single-use tooling.

The global clear aligner market is worth around £5 billion and growing at ~27% a year. Every aligner on the planet is currently made the same way: a one-shot resin mould is 3D printed, used once, and thrown away. A typical patient burns through ~52 of them. A mid-sized lab spends upwards of £400,000 a year on resin that ends up in a skip. We've invented the thing that kills that. There's just one thing standing between the hardware and the world. The software.

The Role:

• Pinpoint is the computational brain of Ghost, our dental machine. It takes a patient's STL dental scan and works out exactly how ~20,000 individually addressable 0.3mm pins should configure themselves to form the mould.
• It validates that the result will actually thermoform a clinically acceptable aligner before a single pin moves.
• You will own the full pipeline, including:

• Geometry engine — STL ingest, mesh repair, surface analysis, curvature and normal computation across meshes up to 500k+ triangles
• Multi-directional pin projection — projecting tens of thousands of pins from multiple tool directions, picking the right pin per region
• Split-line computation — deciding dynamically where coverage hands off between tools so the formed surface is seamless
• Surface mapping & gap resolution — turning a discrete pin field into something that behaves like a continuous mould
• Validation — predicting forming defects (undercuts, bridging, thinning) before they happen
• Compensation engine — an adaptive, data-driven system that learns from real forming results to refine material behaviour models over time
• Operator UI & machine output — a cross-platform desktop app a lab technician can drive a 60-stage case through in under two minutes of manual interaction

One Non-Negotiable: You Build With AI

If you are not actively building with AI — Claude Code, Cursor, Codex, whatever the bleeding edge looks like the week you read this — you are not our candidate. We are looking for someone who treats AI tools as a force multiplier on their thinking, has strong opinions about how to use them well, and is already an order of magnitude more productive than two years ago because of them.

This is not about replacing engineering judgement. The hard parts of Pinpoint — architecture, geometry, trade-offs, taste — are still on you. But the throughput we need from a one-person software function in a hardware company shipping into medical is only achievable if you lean into every tool available.

If your instinct on reading that is excitement, we should talk. If it's defensiveness, we shouldn't.

What We Need From You:

• Must have:

• Deep C++ or Rust in a performance-critical production context
• Computational geometry, mesh processing, or 3D graphics background — not a wrapper-around-a-library job
• Shipped production desktop or embedded software that runs reliably for days
• Comfort owning a system end-to-end: architecture, implementation, performance, the lot
• Judgement to know when to build something clever and when to ship the simplest thing that works

• Strongly preferred:

• GPU compute — CUDA, OpenCL, or compute shaders — on real workloads
• Familiarity with CGAL, libigl, Open3D, or equivalents
• Hands-on with CAD/CAM, slicer software, dental tech, additive manufacturing, or any field where geometry meets physical machines
• Software that talks to hardware — serial protocols, machine controllers, motion systems
• Cross-platform desktop UI work (Qt, Electron, or a strong opinion on what's better)

• Nice to have:

• Differential geometry, optimisation, or applied ML background
• Medical device software experience — MDR, FDA 510(k), IEC 62304
• Built and run a small engineering team before

You do not need a dental background. You do not need a medical background. We will teach you the domain. We need you to be excellent at the engineering.

Equity & Backing:

£3.2m raised. Series A targeted 2027. Three patents pending. Backed by Scott Crump (inventor of FDM, founder of Stratasys) and Al Siblani (EnvisionTEC, $300m exit). Founder Josh Shires previously exited Mous at £56m.

Meaningful equity — all employees own a piece of Fyous. Sheffield HQ — remote considered for the right candidate.