GPU Performance Architect: Low-Latency CUDA Mastery

SIG Susquehanna; Pennsylvania; On-site; USD 120,000 - 150,000; Full time

Job summary

SIG Susquehanna is seeking a GPU Performance Engineer to optimize CUDA kernels for low-latency inference workloads. You will collaborate with quantitative researchers to enhance performance through low-level optimizations while managing complex model structures.

This position demands a PhD in a quantitative field, solid programming skills in C/C++, and a deep understanding of GPU architectures. Join us in shaping optimized compute pipelines that significantly enhance inference performance.

Qualifications

• Strong proficiency in writing and optimizing CUDA kernels.
• Solid programming experience in C/C++.
• Deep understanding of GPU architecture.
• Strong problem-solving skills.

Responsibilities

• Design and optimize custom CUDA kernels.
• Analyze models to identify computational bottlenecks.
• Collaborate with researchers for high-performance pipelines.
• Profile and benchmark GPU performance.
• Contribute to GPU architecture decisions.

Skills

CUDA optimization

C/C++ programming

GPU architecture understanding

Performance analysis

Numerical stability reasoning

Education

PhD in a quantitative field

Tools

ONNX Runtime

TensorRT

Triton

Job description

SIG Susquehanna is seeking a GPU Performance Engineer to optimize CUDA kernels for low-latency inference workloads. You will collaborate with quantitative researchers to enhance performance through low-level optimizations while managing complex model structures.

This position demands a PhD in a quantitative field, solid programming skills in C/C++, and a deep understanding of GPU architectures. Join us in shaping optimized compute pipelines that significantly enhance inference performance.