Build and ship production ML models using modern frameworks (PyTorch preferred) to analyze single-cell multi-omics and functional assay data. You’ll work at a TechBio company using patient tissue data to drive disease understanding from cause to cure.

Xelix★0.0· London

You’ll lead design and delivery of production-grade AI solutions combining machine learning, LLMs, and robust engineering, architecting complex agent-based workflows and tool ecosystems. You will work with modern AI-assisted development tools and define evaluation frameworks. Xelix automates financial controls for enterprise Accounts Payable teams.

Xelix★0.0· London

Work with Python, LLMs (hosted and self-hosted), traditional ML (e.g. gradient boosting), SQL, and pandas to design and build production AI systems for automating Accounts Payable workflows. You will develop agent-based systems, fine-tune models, and write clean, scalable code for the AI Engineering team at this Series-B financial-controls scale-up.

You will work on world models, diffusion models, and multimodal AI, developing novel architectures, training paradigms, and large-scale distributed training/inference infrastructure. The lab creates next-generation foundation models to understand, predict, and interact with the physical world, targeting applications in robotics, autonomous systems, and other domains.

Work with Python, PyTorch, generative models, and cloud infrastructure for ML deployment and optimisation. Join an AI research company developing generative models for scientific discovery, deploying and fine-tuning them for customers.

Build GenAI applications with LLMs and orchestration frameworks using production-grade Python on Azure and Databricks. Apply CI/CD, containerisation, and MLOps to take ML systems from ideation to production for a UK consumer services company’s 15-person central AI function.

Deloitte★3.9 (14k)· London

Work with Python, R, Azure, Databricks, SQL, Tableau, Power BI, Docker, Kubernetes, and other ML frameworks. You will deliver data analytics and machine learning services for Deloitte’s Audit and Assurance practice, providing insights to clients and developing innovative ML tools.

Lead infrastructure and AI infrastructure teams (5-15+ engineers) in London, managing directly and scaling the site from the ground up. Build and operate large-scale AI/ML systems using Kubernetes, Terraform, AWS/GCP/Azure, GPU orchestration, and CI/CD. The company creates AI-powered search and knowledge tools.

You'll work on post-training systems using reinforcement learning, reward modeling, and data generation pipelines to transform pre-trained models into aligned agents. Reflection builds open-weight foundational AI models for various stakeholders, requiring deep LLM training expertise, distributed systems knowledge, and experience improving model behavior through data and RL techniques. You'll collaborate across research and infrastructure teams in a small, high-agency startup environment.

Boehringer Ingelheim★4.0 (2k)· London

Build entity linking and cross-modal integration pipelines using modern data engineering stacks, handling medical imaging (radiology/histopathology) or multi-omics (transcriptomics, proteomics) data within Trusted Research Environments. You will join the AI Enablement team to produce AI-ready, integrated multimodal datasets for model training, fine-tuning and inference.

You'll use Python, Apache Airflow, MLFlow, Docker, Kubernetes, and AWS to build and maintain production ML pipelines, deployment infrastructure, and monitoring for a clinical monitoring platform that detects early signs of clinical deterioration. This role owns the complete ML infrastructure lifecycle, including pipeline orchestration, model serving, experiment tracking, and operational observability.

The role involves designing and delivering AI solutions using Microsoft 365 Copilot, Copilot Studio, Power Platform, Azure AI services, and Google GCP. This position is within SMBC Group’s AI & Innovation team, which shapes and delivers AI strategy, governance, and enterprise AI solutions across the EMEA region.

Design, train, and evaluate deep learning models on petabytes of high-dimensional molecular and clinical data, while building and scaling the core training infrastructure. You’ll be an early technical hire at an AI biology lab that builds world models of immune disease to simulate disease and predict drug response, working within a small, intense team.

You will create machine learning models using PyTorch, NumPy, SciPy, and scikit-learn in areas like player safety, game agents, and world/scene generation. You will work with Epic’s Special Projects & Epic Research Group on applied research at the intersection of vision, language, and speech processing for game development.

Build AI systems end to end using Python, simulators, synthetic data, reinforcement learning, and language-model systems. Aeris-UK develops autonomy software for drones, space debris detection, and cyber defence, deploying models from

Work with Python, LLMs, and distributed training to fine-tune, post-train, and distill language models while building inference systems and data pipelines. This London AI startup builds production ML systems—spanning large-scale modeling and real-world decision support—for major global organizations. You will own the path from ML prototype through deployment, optimization, and scale.

Use Python and PyTorch to build and scale generative and predictive models of cellular behaviour, implementing distributed training, profiling GPU workloads, and optimising inference. Partner with ML Scientists to translate research into production-grade systems applied to multi-modal biological data for drug discovery.

PyTorch, JAX, distributed training (DeepSpeed, FSDP, Ray Train), and Python are used to develop and productionise large-scale biomedical foundation models. The team partners with AI Scientists at a biotechnology company to turn research into scalable ML systems for identifying therapeutic opportunities and accelerating drug discovery.

Working with PyTorch, Hugging Face Transformers, and LLM fine-tuning (LoRA/QLoRA/PEFT), you will develop and deploy embedding, reranking, and semantic search systems for an early-stage FinTech building AI for commodity trading. You will own problem-solving on a small engineering team.

Work with PyTorch Distributed, DeepSpeed, FSDP, Ray Train, MLflow, Weights & Biases, Terraform, and CI/CD for ML workflows. You will own production ML models for an AI-enabled drug discovery organisation, deploying and monitoring models that support therapeutic-target and patient-population decisions.

You'll work across Python, GPU programming (CUDA, JAX XLA), PyTorch or JAX, and finite volume/element methods to build a GPU-native multiphysics simulation platform integrating CFD, CHT, FEA with scientific ML (neural operators, PINNs). The team is developing the full-stack Physics AI infrastructure for engineering, including solver development, physics AI model training, and HPC performance optimisation.

You will use Python to build production-grade data pipelines, transform raw enterprise data into Parquet files, and implement strategies for Continued Pre-Training, Supervised Fine-Tuning, Direct Preference Optimisation, and Reinforcement Learning workflows. As a forward-deployed engineer, you will embed with client teams to design datasets and advise on the complete data lifecycle for domain-adapted LLMs and embedding models. Zettafleet is a platform enabling businesses to train LLMs on proprietary data using non-conventional AI hardware.

Trainline★3.9 (17)· London

Build and productionise machine learning models using Python, Pandas, NumPy, Scikit-learn, Spark, Docker, Terraform, MLFlow, and CI/CD. You will own the end-to-end ML lifecycle for Trainline’s search, recommendations, pricing, personalisation, and generative AI features across the rail-ticketing platform.

You will implement and optimize large neural network models in PyTorch, build infrastructure for distributed training across multi-GPU and multi-node configurations, and develop robust ML pipelines. As part of Relation Therapeutics, you will help build generative and predictive models of cellular behavior to drive drug discovery using single-cell multi-omics and functional assays.

This role involves implementing ML, AI, and LLM research using Python, C++, and deep learning frameworks like TensorFlow, PyTorch, or JAX. You will work on applied and fundamental AI projects focused on semiconductor design and optimisation, building prototypes and translating theoretical models into practical systems.