London · 3 months ago
You will implement and optimise large neural network models and build distributed training and inference infrastructure, partnering with ML scientists to take biology-focused generative and predictive models from prototype to production.
About Relation Relation is a sector defining TechBio company developing transformational medicines, with technology at our core. Our ambition is to understand human biology in unprecedented ways, discovering therapies to treat some of life’s most devastating diseases. We leverage single-cell multi-omics from patient tissue, functional assays, and machine learning to drive disease understanding, from cause to cure. We are scaling rapidly and building a team of exceptional individuals to push the boundaries of drug discovery. You will work in highly interdisciplinary teams where biology, computation, and engineering come together to solve complex problems that have not been solved before. Our state-of-the-art wet and dry labs in the heart of London are designed to accelerate this integration and translate insight into impact. We are committed to building diverse and inclusive teams. Relation is an equal opportunities employer and does not discriminate on the basis of gender, sexual orientation, marital or civil partnership status, gender reassignment, race, colour, nationality, ethnic or national origin, religion or belief, disability, or age. By joining Relation, you will help define how medicines are discovered and deliver meaningful impact for patients. The opportunity Relation is offering an outstanding opportunity for a Machine Learning Research Engineer to help build and scale the next generation of generative and predictive models of cellular behaviour. Research Engineers at Relation are software engineers with a deep understanding of machine learning and deep learning, acting as a critical bridge between theory and implementation: designing, building, and scaling the complex systems on which our ML research depends. You'll be joining a team with access to cutting-edge multiomic and interventional datasets, advanced computational infrastructure, and deep interdisciplinary expertise, and a culture that embraces modern ML tooling, including agentic workflows. Key responsibilities include implementing and optimising large neural network models; building robust infrastructure for distributed training, evaluation, and inference; and partnering with ML Scientists to take research from prototype to production-grade systems applied to large, multi-modal biological data, tested directly in experimental biology. Day to day, you will • Implement and optimise large models, partnering with ML Scientists to translate research ideas into reproducible, scalable training pipelines. • Profile and optimise training across compute, memory, and I/O, pursuing measurable gains in throughput, convergence, and stability. • Design and implement distributed training strategies across multi-GPU and multi-node configurations. • Build and maintain core ML infrastructure. • Contribute to architectural and algorithmic decisions, bringing engineering judgment into research discussions. • Optimise inference and downstream deployment so models can be used by data scientists and biologists in our discovery workflows. • Address numerical, performance, and reliability issues across the stack. • Establish and maintain engineering practices in research code. • Track developments in ML systems and bring relevant advances into our stack. Professionally, you will have • A degree in Computer Science, Engineering, Physics, or a related quantitative discipline; industry experience as an ML / research engineer working on large neural network training. • Strong software engineering fundamentals in Python and deep expertise in PyTorch (or equivalent modern ML frameworks). • Hands-on experience training large neural networks at scale, including distributed training frameworks. • Demonstrable experience profiling and optimising GPU workloads. • Working knowledge of cloud-based ML infrastructure and containerised environments. • A track record of taking research code from prototype to robust, reusable infrastructure that other people actually use. Bonus experience: • CUDA / Triton kernel development; FlashAttention-style attention implementations; experience with foundation models for biology, vision, or language; contributions to open-source ML frameworks. Personally, you • Are comfortable working in a matrixed environment, balancing multiple stakeholders and contributing effectively across teams. • Take ownership of your work, proactively seek opportunities to contribute, and enable others to do their best work. • Communicate openly and directly, give and receive feedback constructively, and handle challenging conversations with respect. • Actively seek out diverse perspectives, build strong working relationships, and contribute to shared goals across teams. • Embrace challenges with openness and resilience, set high standards for yourself, and strive to deliver meaningful outcomes. Working Style & Culture at Relation At Relation, we operate in a matrixed, interdisciplinary environment, where impact is driven through collaboration across scientific, technical, and operational domains. We collaborate, and you will partner with colleagues across multiple teams and projects, contributing your expertise while aligning to shared company priorities. We work together and win together! The patient is waiting! Recruitment Agencies Please note that Relation does not accept unsolicited resumes from agencies. Resumes should not be forwarded to our job aliases or employees. Relation will not be liable for any fees associated with unsolicited CVs.
Posted Aug 10, 2026
