Description
AWS Neuron Collectives is hiring a Software Engineer to optimize collective operations for AWS Trainium, a purpose-built AI training chip. The role focuses on improving collective algorithms and data-center training topologies, analyzing processor, DMA, firmware, and workload metrics, and developing C/C++ implementations of collective communication patterns. The engineer will work closely with hardware teams to maximize compute and bus bandwidth, build analysis frameworks, and investigate improvements for modern LLM training topologies. The position is based in Cupertino, California, and offers a base salary of $165,200 to $223,600 annually.
