GPU Performance Engineer: Scale ML Inference & Systems

Anthropic; New York (NY); Hybrid; USD 280,000 - 850,000; Full time

Job summary

Anthropic is seeking a GPU Performance Engineer to architect and optimize GPU systems powering Claude and related AI models. You will maximize GPU utilization, implement kernel-level optimizations, and scale pipelines across thousands of GPUs.

Ideal candidates will have deep GPU programming experience, a track record of performance improvements, and enjoy solving complex hardware-to-software challenges in a collaborative environment.

Qualifications

• Experience with GPU programming and optimization at scale.
• Ability to navigate complex hardware and ML frameworks.
• Proven track record delivering measurable performance improvements.
• Willingness to collaborate and pair program.

Responsibilities

• Design and optimize GPU kernels for large models.
• Develop distributed multi-node GPU training and inference pipelines.
• Profile and optimize GPU utilization in production ML systems.
• Collaborate with researchers and engineers to push AI infrastructure forward.

Skills

GPU programming

System optimization

Collaborative problem-solving

ML frameworks familiarity

Education

Bachelor’s degree

Tools

CUDA

Triton

CUTLASS

Flash Attention

Nsight

Job description

Anthropic is seeking a GPU Performance Engineer to architect and optimize GPU systems powering Claude and related AI models. You will maximize GPU utilization, implement kernel-level optimizations, and scale pipelines across thousands of GPUs.

Ideal candidates will have deep GPU programming experience, a track record of performance improvements, and enjoy solving complex hardware-to-software challenges in a collaborative environment.