Machine Learning Jobs

• London
• Hybrid
• Senior
• Permanent

Train, debug and improve computer vision and 3D perception models for driver assistance, working across the full ML lifecycle from data and training to evaluation and iteration. The role spans latency-constrained on-car perception and large-scale offline work with lidar, maps and extra compute, building auto-labelling pipelines that feed a data flywheel.

LiDAR

• London
• Hybrid
• Permanent

Build and operate backend services, APIs and data-processing workflows for a platform that deploys, runs and evaluates ML models over petabytes of driving data and large GPU fleets. Work spans model-execution workflows, versioned outputs, annotation and dataset-catalogue features, plus reliability, performance and cost work across distributed, compute-intensive systems.

Flyte · Airflow · Dagster +7

• London
• On-site
• Mid-level · manages people
• Permanent

Join the Preparedness team to anticipate and mitigate loss-of-control risks from increasingly capable AI systems, potentially culminating in recursive self-improvement. The role spans scalable oversight, automated auditing, monitorability testing, model behaviour science, and coordination/verification mechanisms, alternating between rigorous hypothesis-driven research and building interventions that impact production models.

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

Join OpenAI's Training team in London to push the frontier of LLM development for flagship models, enhancing intelligence, efficiency, and adding new capabilities. The role involves designing and scaling new architectures, running experiments, debugging regressions, and optimizing model and computational performance, with a hybrid schedule of three days in the office per week.

• United Kingdom
• Remote
• Senior

Design, develop and optimise GPU/CPU-accelerated numerical libraries such as DALI and nvImageCodec, with heavy focus on performance tuning, benchmarking and CUDA-based parallel algorithms. The role also involves setting technical roadmaps with partners and providing technical leadership and guidance to other library engineers.

CUDA · DALI · nvImageCodec +1

• London
• Hybrid
• Senior
• Permanent

Lead the design, build and deployment of production-grade ML systems for national security and AI safety clients, including capability testing of frontier models. Act as a technical advisor to customers, define deployment standards, and mentor junior engineers while working with frameworks like TensorFlow or PyTorch on cloud and Kubernetes infrastructure.

TensorFlow · PyTorch · AWS +5

• London
• Hybrid
• Mid-level
• Permanent

Builds and deploys production-grade machine learning systems for national security and AI safety clients, including LLM applications, multi-agent tooling and evaluation of frontier models. Involves leading technical scoping and architecture decisions, defining deployment standards, and advising customers while working alongside engineers, data scientists and commercial leads.

AWS · Azure · GCP +5

• United Kingdom
• Hybrid

Designs and builds proactive agentic systems, from sensor-edge reasoning to high-level planning, with stateful multi-turn reasoning, persistent memory and evaluation pipelines. Works across Python, TypeScript, LangGraph, CrewAI, MCP and cloud infrastructure, partnering with biologists and roboticists to take experimental AI from prototype to production.

NextJS · Pydantic · LangGraph +7

• United Kingdom
• Hybrid
• Lead

Senior technical authority architecting algorithmic solutions for intelligent systems, spanning machine learning, applied statistics and sensor processing. Leads complex defence and security projects, sets technical direction, mentors engineers and shapes proposals with clients. Hybrid working, though some projects require full-time office presence.

Reinforcement Learning · Machine Vision · Edge Computing +1

• Exeter
• Entry level
• Contract

A PhD studentship investigating whether repeat photography, citizen science and machine learning can robustly monitor ecological change in intertidal habitats. The researcher combines field data collection, statistical analysis and image-based AI, validating methods against the 25-year MarClim time series across 100 UK rocky intertidal sites.

MarClim

• London
• Hybrid
• Senior

Owns the software engineering that brings data science and machine learning products into live production for the Autoplanner team, building AI-powered planning systems across global warehouses and last-mile assets. Writes typed, well-tested Python (with some Java for OSP integration), builds APIs and data contracts, manages AWS deployments and CI/CD, and packages, serves and monitors ML models. Acts as engineering partner to data scientists and shares on-call support.

AWS · GCP · Ocado Smart Platform

• United Kingdom
• Remote
• Mid-level
• Permanent

Builds and scales data pipelines from ingestion to delivery, implementing dbt best practices, data standards and quality controls. Develops self-service tooling and documentation so teams across Engineering, RevOps and Growth can access data independently, and creates AI agents to answer data questions. Works remotely within a lean, high-velocity data team.

dbt · Python · SQL +2

• London
• Hybrid
• Lead

Lead the applied AI transformation across a legal services group, identifying high-value opportunities and building LLM-powered solutions that reach production. Work with engineers, architects and senior stakeholders to deploy RAG and prompt-engineering solutions, define success metrics and monitor performance. Hybrid role with UK travel.

Azure · Databricks · Snowflake

• Oxford
• On-site
• £40,178 – £48,700 pa
• Entry level
• Contract

Develops and validates machine-learning models on large-scale multimodal NHS clinical data, including electronic health records, genomics, imaging and digital pathology. Builds reproducible pipelines for heterogeneous longitudinal data within secure environments, working with clinicians, pathologists and engineers on disease progression and treatment response prediction.

NHS Data · Electronic Health Records · Secure Data Environment +2

• Norwich
• £20,409 pa
• Entry level
• Contract

Develops and evaluates methods for reliable, localised and generalisable knowledge updating in large language models, adapting techniques such as ROME and MEMIT and comparing them with RAG, external memory and fine-tuning. Experiments use open-weight models on RTX 6000 Ada GPUs and UEA HPC, assessed on benchmarks like CounterFact and zsRE for efficacy, generalisation, locality and retention, with optional deployment in social robots such as Furhat.

ROME · MEMIT · RAG +5