CUDA Kernel Engineer (Remote US)

📋 Description

• Design and optimize custom CUDA kernels for NVIDIA GPUs to maximize occupancy.
• Profile GPU workloads with Nsight Compute, Nsight Systems, nvprof, CUDA-MEMCHECK.
• Eliminate bottlenecks: warp divergence, memory coalescing, PCIe transfers.
• Improve memory pipelines (global/shared/L2/texture) and ensure proper coalescing.
• Collaborate with AI systems, model acceleration, and backend teams.
• Contribute to GPU architecture decisions, libraries, and best practices.

🎯 Requirements

• Proven track record building CUDA kernels from scratch, not just libraries.
• Optimize kernels: tiling, occupancy tuning, shared memory, warp scheduling.
• Deep CUDA knowledge: threads, warps, memory hierarchy, coalescing, warp divergence.
• Experience diagnosing PCIe bottlenecks and host-device transfers.
• Familiar with C++, CUDA runtime APIs, and GPU debugging/profiling tools.
• Experience with multi-GPU systems (NCCL, NVLink, MIG).

🎁 Benefits

• Competitive salary & equity options
• Sign-on bonus
• Health, Dental, and Vision
• 401k