Founding Engineer - ML SystemsUp to £180,000

About the Role

Zenithon AI builds foundation models for extreme physics — the regimes behind semiconductors, aerospace, defence, and fusion energy, where existing simulation tools are too slow, too brittle, or too expensive to keep pace with modern engineering. Their team trains machine learning systems that learn from high-fidelity simulation and experimental data and return fast, reliable predictions with uncertainty estimates. The team is small and highly technical — mostly researchers — and this is the first dedicated engineering owner for the ML stack. You'll own the model code, training and evaluation workflows, experiment infrastructure, and the repo standards that let a research-heavy team move quickly without the codebase becoming fragile. You'll also add distributed-training capability the team doesn't have today, and own the back end of the platform that delivers these models to customers.

This isn't a narrow systems role and it isn't pure research. You'll turn papers and prototypes into clean, well-tested code, integrate independently developed research branches into one coherent codebase, profile and fix the real bottlenecks when training is slow or unstable, and build the in-house tooling — hyperparameter optimisation, ablation, experiment tracking — that makes everyone else faster. There's genuine latitude to contribute to model development itself if that's where your interests run. You'll set the engineering culture for the ML team from the early days.

What They're Looking For

• You're a senior, hands-on engineer (senior / staff / principal / founding / research engineer or ML systems lead) who still writes and ships the critical code yourself.
• You have strong PyTorch experience across model code, data pipelines, training loops, and runtime behaviour — not only at the modelling layer.
• You've done distributed training in practice: multi-GPU or multi-node, GPU clusters, and the memory, communication, and stability problems that come with it.
• You have real model-engineering experience in physics/simulation, vision, or LLM systems, and a track record of improving ML systems through measurement and profiling rather than guesswork.
• You bring strong Python platform and backend foundations — sensible API design, workflow orchestration, and enough Docker / Kubernetes / IaC to build systems others can operate.
• You thrive with high agency and low ego in an early-stage team: you spot what needs doing and do it, rather than waiting for a playbook.

About Zenithon AI

Zenithon AI is building the first foundation models for extreme physics — the physics that underpins semiconductor manufacturing, aerospace, defence, and fusion energy. The mission is to accelerate progress in each of these fields by years: enabling designs that traditional workflows can't reach and helping technologies like fusion energy move toward the grid. It's an early-stage, venture-backed company (seed stage), with a founding team whose backgrounds span fusion energy, plasma physics, and machine learning. Joining now means being one of the first engineers shaping both the core ML systems and the engineering culture.