Machine Learning Performance Engineer

About This Opportunity

We are representing a confidential quantitative trading organization in their search for a Machine Learning Performance Engineer based in London. Our client operates at the intersection of advanced machine learning and high-frequency systems, where performance optimization directly impacts business outcomes.

This role focuses on optimizing ML model performance across both training and inference workloads. You'll work on efficient large-scale training, low-latency inference in real-time systems, and high-throughput inference for research applications. The position requires a whole-systems approach, spanning CUDA optimization, storage systems, networking, and both host- and GPU-level considerations.

Responsibilities

• Optimize performance of machine learning models for both training and inference workflows
• Debug and enhance large-scale distributed GPU training systems end-to-end
• Implement low-latency inference solutions for real-time trading systems
• Design and optimize high-throughput inference pipelines for research environments
• Profile and tune CUDA kernels, memory hierarchies, and GPU compute resources
• Optimize networking and storage infrastructure supporting GPU clusters
• Analyze system performance from L2 cache behavior through cluster-level throughput
• Evaluate and implement new tools and approaches for ML platform efficiency

What We're Looking For

• Strong understanding of modern machine learning techniques and toolsets
• Systems knowledge required to debug training run performance end-to-end
• Low-level GPU expertise including PTX, SASS, warps, cooperative groups, Tensor Cores, and memory hierarchy
• Proficiency with debugging and optimization tools: CUDA GDB, NSight Systems, NSight Compute
• Experience with GPU libraries: Triton, CUTLASS, CUB, Thrust, cuDNN, cuBLAS
• Deep intuition about CUDA performance characteristics including graph launch, tensor core operations, warp synchronization, and asynchronous memory operations
• Knowledge of GPU networking technologies: Infiniband, RoCE, GPUDirect, PXN, rail optimization, NVLink
• Understanding of collective algorithms in NCCL or MPI for distributed training
• Inventive problem-solving approach with willingness to challenge existing methods
• Fluency in English

How We Work With You

As the exclusive recruiting partner for this position, Get A Job.ai manages the entire application process. When you apply through our platform, one of our specialized recruiters will conduct an initial screening to understand your background and technical expertise. We'll then coordinate your submission to our client and guide you through their interview process. Please do not attempt to contact the client directly, as all applications must come through our talent team to be considered.

Pay

Compensation details will be discussed during the screening process with our recruiting team and are competitive for this specialized role in the London market.

Get A Job.ai is committed to providing equal employment opportunities to all qualified individuals regardless of race, color, religion, sex, gender identity, sexual orientation, national origin, age, disability, or veteran status.