GPU Systems Performance Engineer

Janestreet; Greater London; On-site; GBP 111,000 - 140,000; Full time

Job summary

Jane Street in London seeks an engineer with deep experience in low-level systems programming and ML optimisation to join our growing ML team. You will optimise model training and inference, tackling throughput and latency in real-time systems, while addressing storage and GPU-level considerations.

We value an inventive approach and strong collaboration across teams; fluency in English is essential. If you enjoy solving hard technical problems, you’ll fit right in.

Qualifications

• Understanding modern ML techniques and toolsets.
• Experience debugging end-to-end training performance.
• Low-level GPU knowledge and memory hierarchy understanding.
• Familiarity with GPU networking and distributed training.

Responsibilities

• Optimize performance of ML models during training and inference.
• Work on systems-level optimisations including storage, networking, and host/GPU considerations.
• Assess throughput vs goodput and latency implications.
• Collaborate with ML and ML engineering teams in a finance environment.

Skills

ML techniques

End-to-end performance debugging

English fluency

Tools

CUDA

GDB

Nsight Systems

Nsight Compute

Tensor Cores

NCCL

MPI

cuDNN / cuBLAS

Job description

Jane Street in London seeks an engineer with deep experience in low-level systems programming and ML optimisation to join our growing ML team. You will optimise model training and inference, tackling throughput and latency in real-time systems, while addressing storage and GPU-level considerations.

We value an inventive approach and strong collaboration across teams; fluency in English is essential. If you enjoy solving hard technical problems, you’ll fit right in.