Shape the future of Edge AI at Amazon Devices as an ML Kernel Performance Engineer. Focus on crafting efficient CUDA and Triton kernels to optimize neural network compression algorithms for both edge and cloud deployments.
As an ML Kernel Performance Engineer within the Edge AI & Science team, you will work at the intersection of software and hardware. Your expertise will help create high-performance kernels that enhance GPU utilization during training and inference. Collaborate with scientists and engineers to ensure that novel quantization techniques translate into significant throughput improvements on GPUs.
Key Responsibilities:
• Design high-performance CUDA and Triton kernels for AI training
• Analyze kernel performance and optimize compression workloads
• Implement optimizations like operator fusion and scheduling
• Create a kernel development harness for team-wide access
• Collaborate with scientists to enhance training kernels library
Requirements:
•...