Machine Learning Performance Engineer

Trading Interview; New York (NY); On-site; USD 170,000 - 210,000; Full time

Job summary

Jane Street is seeking an engineer with expertise in low-level systems programming and optimisation to join our ML team. The role focuses on optimizing model performance for training and inference, with attention to scalable training, low-latency real-time inference, and high-throughput research workloads.

You will engage with CUDA, GPU memory hierarchies, and distributed systems, ensuring throughput aligns with real throughput expectations.

Qualifications

• Understanding of modern ML techniques and toolsets.
• Experience debugging a training run's performance end to end.
• Low-level GPU knowledge including PTX, SASS, warps, Tensor Cores.
• Familiar with CUDA graph launch, memory hierarchy and synchronization.
• Experience with GPU networking concepts (Infiniband, NVLink, RoCE).

Responsibilities

• Optimize performance of ML models for training and inference.
• Improve throughput in research and real-time systems; reduce latency.
• Work across storage, networking and host/GPU levels for end-to-end efficiency.
• Debug performance bottlenecks using CUDA tools and profiling suites.

Skills

ML techniques

Performance debugging

GPU architecture

Tools

CUDA GDB

NSight Systems

NSight Compute

PTX

SASS

Tensor Cores

Triton

CUTLASS

CUB

Thrust

cuDNN

cuBLAS

Infiniband

NVLink

Job description

We are looking for an engineer with experience in low-level systems programming and optimisation to join our growing ML team.

Machine learning is a critical pillar of Jane Street's global business. Our ever-evolving trading environment serves as a unique, rapid-feedback platform for ML experimentation, allowing us to incorporate new ideas with relatively little friction.

Your part here is optimising the performance of our models – both training and inference. We care about efficient large-scale training, low-latency inference in real-time systems and high-throughput inference in research. Part of this is improving straightforward CUDA, but the interesting part needs a whole-systems approach, including storage systems, networking and host- and GPU-level considerations. Zooming in, we also want to ensure our platform makes sense even at the lowest level – is all that throughput actually goodput? Does loading that vector from the L2 cache really take that long?

If you’ve never thought about a career in finance, you’re in good company. Many of us were in the same position before working here. If you have a curious mind and a passion for solving interesting problems, we have a feeling you’ll fit right in.

There’s no fixed set of skills, but here are some of the things we’re looking for:

• An understanding of modern ML techniques and toolsets
• The experience and systems knowledge required to debug a training run’s performance end to end
• Low-level GPU knowledge of PTX, SASS, warps, cooperative groups, Tensor Cores and the memory hierarchy
• Debugging and optimisation experience using tools like CUDA GDB, NSight Systems, NSight Computesight-systems and nsight-compute
• Library knowledge of Triton, CUTLASS, CUB, Thrust, cuDNN and cuBLAS
• Intuition about the latency and throughput characteristics of CUDA graph launch, tensor core arithmetic, warp-level synchronization and asynchronous memory loads
• Background in Infiniband, RoCE, GPUDirect, PXN, rail optimisation and NVLink, and how to use these networking technologies to link up GPU clusters
• An understanding of the collective algorithms supporting distributed GPU training in NCCL or MPI
• An inventive approach and the willingness to ask hard questions about whether we're taking the right approaches and using the right tools

We were founded by a small group of traders and technologists in a tiny New York office. Today, we have more than 2,000 employees across five global offices. We trade…