ML Performance Engineer: Scale GPU/CPU ML Workloads

G-Research

Greater London

Hybrid

GBP 90,000 - 150,000

Full time

3 days ago

Be an early applicant

Get more replies from employers

Send a job-specific resume in minutes.

Benefits offered by this job

Competitive pay

Lunch provided

Annual leave 35d

Pension 9%

Informal dress code

Healthcare & life assurance

Cycle-to-work

Monthly events

Job summary

G-Research in London seeks an exceptional ML Performance Engineer to optimise large-scale workloads across GPU and CPU infrastructure. You will profile and tune training and inference jobs, develop reference implementations, and collaborate with research and platform teams to evolve the compute stack.

The role requires strong Python, CUDA, and PyTorch knowledge, plus experience with HPC schedulers and Kubernetes.

Qualifications

Proven track record profiling, benchmarking and optimising distributed workloads.
Experience with Python.
Knowledge of CUDA.
Experience with HPC schedulers and Kubernetes-based workload orchestration.
Strong understanding of PyTorch and deep learning frameworks.
Strong background in data structures, algorithms and parallel programming.
Deep understanding of Linux OS fundamentals.
Familiarity with profiling/monitoring tools (nsys, ncu, eBPF).
Strong communication skills across teams.

Responsibilities

Collaborate with researchers and engineers to understand compute challenges and design optimised solutions.
Profile, benchmark and tune large-scale training and inference workloads on distributed CPU, GPU, and memory-intensive jobs.
Develop reference implementations, libraries and tools to improve job efficiency and reliability.
Collaborate with systems, architecture and platform teams to evolve the compute stack.
Influence long-term platform and infrastructure decisions.