Software Engineer (Kernels/CUDA, C++)

• We are building one of the world’s largest AI supercomputers from the ground up. As part of the Compute Infrastructure team, you will own both the raw GPU supercomputer and the platform layer that runs on top of it
• You will work across the full stack — from low-level GPU kernel optimizations and Linux kernel internals to massive-scale orchestration and virtualization — to make training and inference at SpaceXAI as fast, reliable, and scalable as possible
• This is a broad, high-impact role that combines hardcore supercompute and compute infrastructure work. Your contributions will directly accelerate Grok’s training speed and overall AI progress
• Design, build, and optimize massive GPU clusters for extreme-scale training and inference workloads
• Develop and tune low-level CUDA kernels (GeMM, Attention, etc.), using CUTLASS, Tensor Cores, and Nsight for maximum performance
• Profile, debug, and eliminate bottlenecks across GPU memory hierarchy, networking fabric, filesystems, and multi-GPU operation
• Collaborate closely with AI research teams to deliver production-grade performance and scalability

Benefits

• Health and wellness: Comprehensive health insurance including medical, dental, vision, and disability coverage
• Life and family: Life and AD&D insurance and fertility benefits to ensure our team’s well-being and peace of mind
• Flexible vacation: We work hard but avoid burn out. Take time off when you need it
• Visa sponsorship: We support international talent with visa sponsorship to join our team
• 401(k) plan: Retirement savings plan to secure your financial future

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