Founding Engineer - ML Systems (up to £180k)

Dex; Greater London; On-site; GBP 120,000 - 160,000; Full time

Job summary

Dex partners with a client that builds foundation models for physics and engineering. You’ll be the first dedicated engineering owner for the ML stack, turning research prototypes into robust code and delivering the backend tooling for customers.

You will own model code, training workflows, and distributed training, integrating research branches into a production-ready codebase while improving performance and scalability with PyTorch, Python, Docker, and Kubernetes.

Qualifications

• Senior hands-on engineer who still writes and ships critical code.
• Strong PyTorch experience across model code, data pipelines, and training loops.
• Proven distributed training with multi-GPU/node setups and memory/communication considerations.
• Model-engineering in physics/ simulation, vision, or LLMs with data-driven improvements.
• Solid Python backend foundations: API design, orchestration, Docker/Kubernetes/IaC.

Responsibilities

• Own the entire ML stack: model code, training/evaluation workflows, and experiment infra.
• Build/distribute training capabilities for large-scale model development.
• Integrate research branches into a production-ready codebase.
• Profile bottlenecks to improve training stability and performance.
• Design and build the backend platform delivering models to customers.

Skills

Senior hands-on engineer

PyTorch

Distributed training

Python backend

Docker

Kubernetes

IaC

Tools

Docker

Kubernetes

IaC

Job description

This role is with one of Dex's trusted partner companies. We work closely with their teams to truly understand their culture, goals, and what they're looking for, so we can match you with the right opportunity and give you context about the role before you commit to a process.

The role

This company builds foundation models for extreme physics: the regimes behind semiconductors, aerospace, defence, and fusion energy. Their work accelerates progress in fields where existing simulation tools are too slow or brittle, enabling designs traditional workflows can't reach. They're backed by leading investors and tackling problems with global impact.

You’ll be the first dedicated engineering owner for the ML stack, joining a small, highly technical team of researchers. This isn't a narrow systems role, nor is it pure research; you’ll turn prototypes into robust code, integrate research branches, and build the in-house tooling for experiment tracking and hyperparameter optimisation. You’ll also add distributed training capabilities and own the backend platform delivering these models to customers, setting the engineering culture from day one.

The work

• Own the entire ML stack: model code, training and evaluation workflows, and experiment infrastructure.
• Build and implement distributed training capabilities for large-scale model development.
• Integrate independently developed research branches into a coherent, production-ready codebase.
• Profile and resolve real bottlenecks in training stability and performance, improving system efficiency.
• Design and build the backend platform that delivers these foundation models to customers.

What You Bring

• You are a senior, hands-on engineer who still writes and ships critical code.
• Strong practical experience with PyTorch across model code, data pipelines, and training loops.
• Proven track record with distributed training (multi-GPU/node, GPU clusters), understanding associated memory and communication challenges.
• Real model-engineering experience in physics/simulation, vision, or LLM systems, with a focus on data-driven system improvement.
• Solid Python platform and backend foundations: API design, workflow orchestration, and practical Docker/Kubernetes/IaC.