Sr. Software Engineer, Kernels/CUDA (C++)

xAI•Seattle, WA

19h

About The Position

SpaceXSI is building Super Intelligence and developing a full stack that spans supercomputers, Grok models, and applications. As part of the Compute Infrastructure team, the Sr. Software Engineer, Kernels/CUDA (C++) will help build and operate a large AI supercomputer, owning both the GPU supercomputer and its platform layer. The role spans low-level GPU kernel optimization and Linux kernel internals through large-scale orchestration and virtualization, with the goal of making AI training and inference faster, more reliable, and more scalable. Work will directly support the performance of Grok training and broader AI development.

Requirements

At least 5 years of experience with low-level system programming in C/C++.

Nice To Haves

Deep low-level systems programming experience with C, C++, PTX, or SASS.
Strong experience with large-scale GPU clusters or distributed compute infrastructure at production scale.
Hands-on experience with GPU kernel optimization, including CUTLASS, custom kernels, and Nsight profiling.
A track record of building or running high-performance infrastructure for AI training or inference workloads.
Ability to reason from first principles and optimize both memory-bound and compute-bound scenarios.

Responsibilities

Design, build, and optimize massive GPU clusters for extreme-scale training and inference workloads.
Develop and tune low-level CUDA kernels, including GeMM and Attention, using CUTLASS, Tensor Cores, and Nsight.
Profile and debug performance, and eliminate bottlenecks across GPU memory hierarchy, networking fabric, filesystems, and multi-GPU operation.
Collaborate with AI research teams to deliver production-grade performance and scalability.