Senior GPU Kernel Engineer — Geometric ML

NVIDIA Corporation; Santa Clara (CA); Presencial; USD 184 000 - 357 000; Tempo integral

Resumo da oferta

NVIDIA Corporation in Santa Clara, CA seeks a Senior Software Engineer for the cuEquivariance team to advance GPU-accelerated ML primitives and production libraries. You will implement CUDA kernels and maintain interfaces for PyTorch and JAX, ensuring scalable, production-ready software across NVIDIA GPU generations.

The role requires strong C++/Python, deep GPU knowledge, and collaboration with research teams and external partners.

Qualificações

• 6+ years of software engineering with CUDA/GPU
• Strong C++ and Python
• Solid foundation in GPU memory hierarchy and profiling
• Experience with production libraries used by external developers
• Familiarity with geometric ML concepts such as equivariance or tensor products

Responsabilidades

• Build, optimize CUDA kernels for equivariant neural networks.
• Deliver end-to-end GPU-accelerated primitives to production.
• Develop PyTorch and JAX interfaces for cuEquivariance primitives.
• Drive CI/CD for multi-GPU kernel builds and testing.
• Collaborate with Applied Science and research teams; interface with third-party developers.

Conhecimentos

CUDA

GPU programming

C++

Python

Geometric ML

Production libraries

CI/CD

Formação académica

BS/MS in CS/Physics/Applied Math

Ferramentas

PyTorch

JAX

Triton

Descrição da oferta de emprego

NVIDIA Corporation in Santa Clara, CA seeks a Senior Software Engineer for the cuEquivariance team to advance GPU-accelerated ML primitives and production libraries. You will implement CUDA kernels and maintain interfaces for PyTorch and JAX, ensuring scalable, production-ready software across NVIDIA GPU generations.

The role requires strong C++/Python, deep GPU knowledge, and collaboration with research teams and external partners.