PhysicsX builds AI software for engineering and manufacturing. Its models help teams simulate complex physical systems faster across industries such as aerospace, energy, semiconductors, materials and automotive. The company raised $300 million in Series C funding in June 2026 at a valuation of about $2.4 billion.
- Build and scale machine learning models for engineering and physics problems
- Work with distributed training, GPU computing and scientific software
- Join a London AI company that raised $300 million in June 2026
The role
PhysicsX is looking for a Machine Learning Software Engineer to help turn scientific research into robust, reusable software. You will work closely with research scientists and simulation engineers on real-world physics and engineering problems.
A major part of the job is scaling models beyond early prototypes. That means improving performance, building distributed training systems and making experiments easier to run and reproduce.
The work includes:
- Building and optimising machine learning models for scientific applications
- Turning research prototypes into reliable implementations
- Developing multi-node and multi-GPU training systems
- Helping train and scale foundation models for science and engineering
- Selecting libraries, frameworks and infrastructure for model development
- Creating reusable tools, pipelines and product components
- Explaining results to colleagues and customers working on practical engineering problems
Who PhysicsX is looking for
You should have an MSc or PhD in a relevant technical subject, or equivalent practical experience. Useful backgrounds include computer science, machine learning, applied mathematics, statistics, physics, engineering and scientific software.
The team is particularly interested in experience with scientific computing, high-performance computing and distributed model training. Relevant tools include Python, PyTorch, JAX, NumPy, SciPy, CUDA, Triton, Spark, Dask, Docker, Kubernetes and Slurm.
Experience with C or C++ for scientific computing, computer vision or geometry processing would also be useful.
Why apply
This role should appeal to an engineer who wants machine learning work tied to physical products and industrial problems. The company is applying AI to areas where simulation can be slow and expensive, so better models can change how quickly engineers test new ideas.
PhysicsX offers equity options, a 10% employer pension contribution, 25 days of annual leave, private medical insurance, free office lunches and enhanced parental leave. Its hybrid model combines time in the Shoreditch office with work-from-home days.