Machine Learning Performance Engineer – Global Trading Firm

Job Overview:

Our client, a global leader in quantitative trading, is seeking a Machine Learning Performance Engineer to join its New York engineering team. This role plays a critical part in advancing large-scale deep learning infrastructure that supports high-frequency and futures trading strategies.

Key Responsibilities

• Design, optimize, and deploy scalable training and inference pipelines for deep learning models.
• Enhance and extend open-source ML frameworks to improve performance and reliability.
• Identify and resolve GPU and system-level bottlenecks for accelerated model training.
• Collaborate closely with researchers, quants, and software engineers to integrate ML models into trading systems.
• Develop a deep understanding of trading workflows, market data, and performance-critical systems.

Qualifications

• Strong expertise in GPU programming (CUDA)—Tensor Cores, cooperative groups, graphs, and warp-level intrinsics.
• Proven experience with deep learning internals in PyTorch, TensorFlow, or JAX.
• Deep understanding of computer architecture, parallel computing, and large-scale distributed training.
• Proficiency in C++ and Python.

Preferred: Experience with JAX ecosystem (XLA, Flax), GPU libraries (Triton, cuDNN, cuBLAS), Linux system programming, and open-source ML contributions.

Why Apply

Join a team of world-class engineers and researchers tackling some of the most advanced challenges in trading and AI infrastructure. This is a greenfield opportunity to shape GPU- and TPU-based deep learning pipelines at scale within a high-impact, globally recognized trading firm.