Machine Learning Jobs

• London
• Hybrid
• £370,000 – £630,000 pa
• Mid-level

Builds and improves the RL training infrastructure researchers rely on, profiling, debugging and rearchitecting bottlenecks across the stack. Partners with researchers and adjacent engineering teams to ship tooling that speeds up training runs, owning reliability and performance end-to-end.

JAX · PyTorch

• London
• On-site
• £260,000 – £630,000 pa
• Mid-level

Own critical parts of a production pretraining pipeline, covering performance optimization, observability, reliability and hardware debugging across the full stack. The role splits roughly 50/50 between research and engineering, with on-call incident response during model launches and close collaboration with teams in SF and London.

JAX · TPU · PyTorch +3

• London
• Hybrid · 3 days in office
• Mid-level
• Permanent

Partner directly with Digital Native customers to design, build and deploy AI-powered products, taking ideas from early experimentation through evaluation, production launch and scale. You will write and debug code, build evaluation systems, resolve complex integrations, and guide decisions on model behavior, reliability, latency, cost, safety and governance.

OpenAI

• London
• Hybrid
• Senior
• Permanent

Builds data-intensive, AI-powered applications and infrastructure that help operators and investigators detect, triage and review high-severity harm, working across UIs, services and pipelines. Partners closely with the Child Safety team and cross-functional groups, owning systems end-to-end in an ambiguous, fast-moving environment.

Spark · Kafka · Flink +3

• London
• Hybrid
• Lead · manages people
• Permanent

Lead a London-based team building and operating large-scale multi-cloud, multi-region GPU compute and storage infrastructure for ML training and inference. The role is people-first: coaching engineers, hiring, and owning roadmap and delivery, while guiding Kubernetes scheduling, capacity planning and cost trade-offs across heterogeneous GPU fleets.

Kubernetes · Kueue · Nvidia Hopper +2

• London
• Hybrid
• Senior
• Permanent

Hands-on optimisation of large transformer model inference for edge accelerators and GPUs, profiling model graphs, compiler/runtime behaviour, kernel execution and memory movement. You'll implement fusions, scheduling, quantisation-aware performance and custom kernels, and build benchmarking and regression tests for targets like NVIDIA Orin/Thor and Qualcomm platforms, working closely with model, platform and deployment teams.

TensorRT · CUDA · Qualcomm QNN +6

• London
• Hybrid · 3 days in office
• Mid-level

Join a newly formed Scheduling team to own and extend systems that orchestrate compute allocation for complex ML workloads on Kubernetes. You'll design scheduling policies like gang scheduling and topology-aware placement, build custom Kubernetes controllers and CRDs, and partner with ML researchers to optimize GPU and TPU utilization.

Kubernetes · Kueue · Volcano +5

• London
• Hybrid · 3 days in office
• Senior

Lead the hardening and scaling of an inference platform serving fleets of cutting-edge ML models for scientific applications. Set strategy and roadmaps, architect next-generation services, and solve core scaling limits for high-throughput performance, partnering with research, product and operations teams.

Kubernetes · Google Cloud Platform · NVIDIA GPU +1

• London
• Hybrid
• £65,000 – £145,000 pa

Engineers at all levels build and scale the shared tooling and infrastructure behind the institute's AI safety research, from Inspect plugins and large-scale evaluation systems to model hosting and secure proxy infrastructure. Work spans open-source collaboration, supporting research teams with custom setups, and hardening infrastructure for frontier model evaluations.

Inspect

• London
• Hybrid
• Senior
• Permanent

Architect novel generative models to design functional proteins, working in an interdisciplinary team of ML researchers, protein engineers and biologists. The role spans data curation, evaluation metrics, rapid prototyping, cloud training and inference optimisation, plus planning wet lab testing campaigns with the bio team.

• London
• Director · manages people
• Permanent

Lead the platform engineering team owning hybrid on-prem and cloud compute, Kubernetes clusters, and MLOps infrastructure such as Kubeflow notebooks, GPU training and model serving. Set architectural direction, mature CI/CD, observability and security controls, and ensure scientific compute meets the needs of ML and data science teams. Reports to the VP of Engineering in a matrixed TechBio environment.

Kubernetes · Kubeflow · ISO 27001 +2

• London
• Senior

Optimise machine learning model performance across training and inference in a tech-centric prop trading fund. Work spans CUDA and GPU-level tuning, low-latency real-time inference, high-throughput research inference, plus storage, networking and host-level systems concerns within a growing ML team.

CUDA · PTX · SASS +7

• London
• Hybrid
• £65,000 – £145,000 pa

The Chem Bio team researches, evaluates and mitigates chemical and biological risks from frontier AI systems, running capability evaluations, building benchmarks and datasets, and developing safeguards and monitoring methods. Findings are shared with frontier AI companies, UK officials and allied governments to inform deployment and policy. The role is open to research engineers and scientists at all seniority levels, with appointment conditional on obtaining UK Government Developed Vetting clearance.