Michael Andersch is a principal GPU architect and senior architecture manager at NVIDIA. He started his career in the Compute Architecture team, where he focused on advancing the GPU's capabilities for the world's diverse set of CUDA workloads. In recent years, Michael has driven both hardware and software improvements specifically to increase the performance of deep neural network training, and his team is now involved wherever training performance is critical; from tuning the latest neural networks in the deep learning community to designing next-generation GPUs.