CUDA Kernel Architect for Low-Latency GPU Inference

Trading Interview; Northern (KY); Hybrid; USD 140,000 - 210,000; Full time

Job summary

Susquehanna is seeking a GPU Performance Engineer to build highly optimized CUDA kernels for low-latency inference. This role targets workloads where standard runtimes fall short, demanding custom kernels, memory layouts, and execution strategies to gain meaningful performance.

You will partner with quantitative researchers to identify bottlenecks, translate models into efficient GPU implementations, and push end-to-end latency improvements in production systems.

Qualifications

• Expert in writing and optimizing CUDA kernels.
• Strong C/C++ programming experience.
• Deep understanding of GPU architecture and memory hierarchy.
• Ability to reason about numerical stability and hardware efficiency.
• Experience with low-level systems and performance analysis.

Responsibilities

• Design, implement, and optimize custom CUDA kernels for latency-critical inference workloads.
• Develop fine-grained GPU implementations tailored to model structures.
• Analyze models to identify bottlenecks and opportunities for parallelization.
• Collaborate with quantitative researchers to translate models into high-performance pipelines.
• Profile, benchmark, and improve end-to-end inference latency and throughput.

Skills

CUDA kernels

C/C++

GPU architecture

numerical stability

low-level systems

Job description

Susquehanna is seeking a GPU Performance Engineer to build highly optimized CUDA kernels for low-latency inference. This role targets workloads where standard runtimes fall short, demanding custom kernels, memory layouts, and execution strategies to gain meaningful performance.

You will partner with quantitative researchers to identify bottlenecks, translate models into efficient GPU implementations, and push end-to-end latency improvements in production systems.