Senior Software Engineer — cuEquivariance

What you'll do

• Implement and optimize CUDA kernels for equivariant neural-network primitives and deliver validated production software.
• Build and maintain PyTorch and JAX interfaces; drive CI/CD, correctness testing, and performance regression tracking.
• Collaborate with research teams and external framework developers to evaluate architectures and integrate software into production pipelines.

What they ask for

• 6+ years of software engineering experience with a strong background in CUDA and GPU programming.
• Deep proficiency in C++ and Python, including building and shipping production libraries for external developers.
• Knowledge of GPU computing fundamentals and performance profiling; experience with production scientific software libraries, ML frameworks, or GPU APIs.
• Familiarity with geometric machine-learning concepts such as equivariance, group representations, irreducible representations, or tensor products.

Nice to have

• Experience with e3nn, MACE, NequIP, SE(3)-Transformers, or similar; Triton or other GPU kernel tools; or mixed-precision and tensor-core-aware design.
• PhD or equivalent experience focused on geometric deep learning or HPC; contributions to open-source geometric ML or GPU computing projects.

Pay

184,000 USD - 287,500 USD base salary (Level 4); 224,000 USD - 356,500 USD base salary (Level 5)

Tools

CUDA, C++, Python, PyTorch, JAX, Triton, e3nn, MACE

AI focus

Develop GPU kernels and software interfaces that accelerate equivariant deep learning; collaborate on evaluating equivariant architectures. NVIDIA states it uses AI tools in recruiting.