Machine Learning Engineer — GPU Kernel

What you'll do

• Profile and optimize deep-learning workloads; identify system bottlenecks and improve GPU utilization.
• Design performance benchmarks and testing methods, and build tools to automate workload analysis and optimization.
• Develop kernels for new model architectures and validate kernel outputs and gradients against reference implementations.

What they ask for

• Strong C++ skills and hands-on CUDA kernel development and optimization for deep-learning workloads.
• Understanding of GPU memory hierarchy, warp/block execution, and compute-memory trade-offs, with profiling-driven optimization experience.
• Strong Python skills and experience integrating kernels with PyTorch or an equivalent framework, including numerical and gradient validation.

Nice to have

• Experience with Triton, CUTLASS, or PTX/SASS analysis.
• Experience with multi-node distributed training or inference systems; mixed-precision validation such as BF16 or FP8.

Pay

USD 150,000–450,000 per year (base salary)

Tools

C++, CUDA, Python, PyTorch, Triton, CUTLASS, PTX/SASS, BF16

AI focus

Optimize machine-learning software and GPU kernels for foundation-model training and inference, and support kernels and systems for new model architectures and algorithms.