Senior GPU Sparse Math Libraries Engineer

NVIDIA Corporation; Santa Clara (CA); On-site; USD 184,000 - 357,000; Full time

Job summary

NVIDIA in Santa Clara is seeking software engineers to design and implement libraries and tools to accelerate computing for unstructured sparsity in DL and HPC. You will work across C++, Python, and CUDA, and contribute to DSL specifications and on-demand code generation for DL frameworks like PyTorch.

You will collaborate with product management and global teams to define features, optimize performance, and deliver production-ready code for scalable GPU-accelerated software.

Qualifications

• 6+ years of experience developing, debugging, and optimizing high-performance software using C++ and parallel programming.
• Experience with sparse linear algebra applications and CUDA, MPI, OpenMP or equivalent technologies.
• Experience with domain-specific language design and compiler optimizations (sparse compilers such as MLIR or TACO).

Responsibilities

• Design and develop a C++-based system to simplify and accelerate computing for unstructured sparsity in DL and HPC on NVIDIA GPUs.
• Enable the system in languages and frameworks used in DL, such as Python and PyTorch.
• Evaluate and improve the performance of the system on real-life applications.
• Write effective and well-tested code for production use to improve library quality, performance and maintainability.
• Collaborate with product management and external partners to understand features and contribute to technical roadmaps.

Skills

C++

Python

CUDA

MPI

OpenMP

High-performance computing

Education

BS/MS/PhD in CS/Applied Math/related field

Tools

MLIR

TACO

Job description

NVIDIA in Santa Clara is seeking software engineers to design and implement libraries and tools to accelerate computing for unstructured sparsity in DL and HPC. You will work across C++, Python, and CUDA, and contribute to DSL specifications and on-demand code generation for DL frameworks like PyTorch.

You will collaborate with product management and global teams to define features, optimize performance, and deliver production-ready code for scalable GPU-accelerated software.