Machine Learning Performance Engineer

3 Months ago • All levels

Research Development

Job Description

We are seeking an engineer with expertise in low-level systems programming and optimization to join our expanding ML team. Your role will focus on enhancing the performance of our machine learning models, covering both training and inference. This includes optimizing large-scale training, achieving low-latency inference in real-time systems, and high-throughput inference in research. While basic CUDA improvements are involved, the core challenge lies in a holistic systems approach, addressing storage systems, networking, and both host and GPU-level considerations to ensure optimal efficiency and performance.

Good To Have:

Experience with Tensor Cores
Knowledge of Infiniband, RoCE, GPUDirect
Insight into CUDA graph launch latency/throughput

Must Have:

Low-level GPU knowledge (PTX, SASS, warps)
Debugging/optimization with CUDA GDB, NSight
Library knowledge (Triton, CUTLASS, CUB)
Understanding of ML techniques and toolsets
Systems knowledge for end-to-end performance debugging
Understanding of collective algorithms (NCCL, MPI)
Fluency in English

Add these skills to join the top 1% applicants for this job

problem-solving

data-structures

cuda

networking

sass

algorithms

machine-learning

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
Fluency in English