ML Performance Engineer — Scalable DL Pipelines & Optimization

Optiver US LLC; New York (NY); On-site; Full time

Benefits offered by this job

Competitive compensation package

Global profit-sharing pool

401(k) match up to 50%

Comprehensive health, dental, vision coverage

Extensive office perks, including food and social events

Job summary

A global trading firm is seeking a Machine Learning Performance Engineer to focus on AI infrastructure. This role involves building training and inference pipelines, enhancing deep learning frameworks, and tackling performance bottlenecks. Candidates must possess strong GPU programming knowledge with CUDA and expertise in frameworks like PyTorch or TensorFlow. Competitive compensation and extensive benefits, including a profit-sharing pool, are offered in this high-performing environment.

Qualifications

• Strong knowledge of low-level GPU programming with CUDA, including Tensor Cores.
• Expertise in internals of deep-learning frameworks like PyTorch, JAX, TensorFlow.
• Deep understanding of computer architecture.
• Experience in C++ and Python.

Responsibilities

• Build scalable and robust training and inference pipelines for deep learning.
• Enhance functionality of open-source deep learning frameworks.
• Identify and eliminate performance bottlenecks.
• Collaborate closely with researchers and other engineers.
• Develop deep understanding of trading systems.

Skills

Low-level GPU programming with CUDA

Deep learning frameworks expertise (PyTorch, JAX, TensorFlow)

Understanding of computer architecture

Experience in C++

Experience in Python

Tools

CUDA

JAX ecosystem

GPU libraries (Triton, CUB, cuDNN, cuBLAS)

Job description

A global trading firm is seeking a Machine Learning Performance Engineer to focus on AI infrastructure. This role involves building training and inference pipelines, enhancing deep learning frameworks, and tackling performance bottlenecks. Candidates must possess strong GPU programming knowledge with CUDA and expertise in frameworks like PyTorch or TensorFlow. Competitive compensation and extensive benefits, including a profit-sharing pool, are offered in this high-performing environment.