AI Platform Engineer

About Revolut People deserve more from their money. More visibility, more control, and more freedom. Since 2015, Revolut has been on a mission to deliver just that. Our powerhouse of products — including spending, saving, investing, exchanging, travelling, and more — help our 80+ million customers get more from their money every day. As we continue our lightning-fast growth, 2 things are essential to our success: our people and our culture. In recognition of our outstanding employee experience, we've been certified as a Great Place to Work™. So far, we have 13,000+ people working around the world, from our offices and remotely, to help us achieve our mission. And we're looking for more brilliant people. People who love building great products, redefining success, and turning the complexity of a chaotic world into the simplicity of a beautiful solution. About the role Our Technology team builds the systems and experiences that keep Revolut moving. From the infrastructure behind our innovative app to the features used by millions of people around the world, they bring sharp thinking, speed, and a focus on meaningful impact to everything they do. We’re looking for a Python Engineer who can write high-quality code and build automated solutions for heavily regulated financial systems. As part of the AI team that's driving innovation and powering products, you'll be developing the core infrastructure behind our central platform that empowers our scientists and engineers to solve tomorrow's challenges today. Up to shape what's next in finance? Let's get in touch. What you'll be doing - Designing, building, and maintaining scalable backend services and tooling that support the AI lifecycle and GPU-based workloads - Profiling and optimising model training and inference workloads (latency, throughput, stability) using PyTorch and existing CUDA-enabled libraries - Improving GPU compute utilisation, memory consumption, and bandwidth across shared, multi-tenant, and multi-node environments - Implementing mixed-precision execution, quantisation, efficient batching, gradient accumulation, and memory-efficient attention - Creating repeatable benchmarking practices, automated regression tests, and acting as a guide for engineering teams selecting execution frameworks and GPU types What you'll need - A track record of designing and operating scalable backend systems in production environments (Linux, Kubernetes, and containerised workloads), primarily using Python - Experience running, profiling, and optimising machine learning workloads (built with PyTorch or equivalent) on NVIDIA GPUs - A solid understanding of GPU execution concepts, including memory hierarchy, CPU/GPU synchronisation, host-to-device transfers, and CUDA stream behaviour - Familiarity with multi-GPU communication (NCCL), data/tensor parallelism, and resource allocation in Kubernetes - A degree in a STEM subject (or equivalent) with a solid foundation in computer science principles, testing, observability, and operational ownership