Senior Ascend Npu Systems Engineer

Advance Ascend NPU performance at Huawei Canada through kernel development and distributed training optimization for foundation models.As a Senior System Engineer, you will develop compute operators and scale pre-training and post-training workloads. The work spans AscendC, Triton, TileLang, MindSpeed, Megatron, and PyTorch FSDP. You will improve performance across Ascend NPU clusters.Key Responsibilities:
Design and tune FlashAttention and GEMM operators using AscendCOptimize MindSpeed pre-training and reinforcement learning post-training pipelinesTune hybrid parallelism and rollout actor pipeliningImplement FP8, MXFP8, and mixed-precision training schemesProfile Ascend hardware using msprof and PyTorch ProfilerRequirements:
Bachelor’s, Master’s, or PhD in a related technical disciplineThree or more years in HPC or distributed model systemsExperience with AscendC, CANN, or CUDA C/C++Proficiency in modern Python and PyTorch custom extensionsExperience with AI hardware profiling and debugging toolsApply AscendC, distributed systems, and performance analysis expertise to foundation model acceleration at Huawei Canada.
#J-18808-Ljbffr