GPU Kernel Engineer — High-Performance ML & Acceleration

7Seventy; Northern (KY); On-site; USD 150,000 - 190,000; Full time

Job summary

7Seventy is seeking a Machine Learning Engineer - Kernels focused on custom GPU and accelerator kernel development for high-performance AI workloads. You will design low-level implementations, optimize performance, and translate advances in ML into production-ready code.

The role blends GPU programming, parallel computing, hardware-aware optimization, and ML infrastructure, with work spanning C++, CUDA, performance profiling, benchmarking, and optimization in distributed and heterogeneous

Qualifications

• 2+ years of GPU programming or parallel computing.
• Experience optimizing workloads for distributed/heterogeneous computing.
• Strong C++ and CUDA programming skills.
• Familiarity with ML backends and profiling tools.

Responsibilities

• Design, develop, and implement custom GPU/accelerator kernels for ML workloads.
• Profile and benchmark ML workloads; identify bottlenecks; implement optimizations.
• Collaborate with ML researchers to translate advances into production-ready code.
• Evaluate accelerator hardware and programming tech (CUDA, ROCm, TPUs) to guide kernel design.
• Document optimization techniques and share best practices across teams.

Skills

GPU kernel development

Parallel computing

Performance profiling

Distributed computing

Hardware acceleration

C++ programming

Low-level optimization

Collaborating with researchers

Education

Bachelor's degree in Computer Science or Electrical Engineering

Master's degree or PhD preferred

Tools

CUDA

ROCm

TPUs

Job description

7Seventy is seeking a Machine Learning Engineer - Kernels focused on custom GPU and accelerator kernel development for high-performance AI workloads. You will design low-level implementations, optimize performance, and translate advances in ML into production-ready code.

The role blends GPU programming, parallel computing, hardware-aware optimization, and ML infrastructure, with work spanning C++, CUDA, performance profiling, benchmarking, and optimization in distributed and heterogeneous