Machine Learning Performance Engineer

Responsibilities

• Optimize ML model training and inference performance at scale
• Improve efficient large-scale training and low-latency real-time inference
• Apply whole-systems thinking across storage, networking, host and GPU levels
• Debug low-level performance issues including memory hierarchy and cache behavior

Requirements

• Understanding of modern ML techniques and toolsets
• Experience debugging end-to-end training run performance
• Low-level GPU knowledge: PTX, SASS, warps, cooperative groups, Tensor Cores
• CUDA debugging tools: CUDA GDB, NSight Systems, NSight Compute
• GPU libraries: Triton, CUTLASS, CUB, Thrust, cuDNN, cuBLAS
• CUDA performance intuition: graph launch, tensor core arithmetic, warp sync, async memory loads
• Networking technologies: Infiniband, RoCE, GPUDirect, PXN, rail optimisation, NVLink
• Distributed training: NCCL or MPI collective algorithms
• Fluent in English

About the Company

Jane Street is a quantitative trading firm where machine learning is a critical pillar of the global business. The firm’s evolving trading environment serves as a rapid-feedback platform for ML experimentation.