Getting the AI you want on the infrastructure you know: 2 deep-dive case studies of AI on CPU

Kushal Datta is a senior research scientist in the Artificial Intelligence Products Group at Intel. He specializes in accelerating deep learning training and inference on Intel architecture. His noteworthy achievements are the optimization of multiscale CNN training time on Multi-Node Xeon and INT8/VNNI quantization of the transformer model. Previously, he contributed to Intel Graph Analytics Toolkit, Genomics Analytics Toolkit-4.0 (from Broad Institute), and TileDB (a multidimensional array store). He earned his PhD in ECE from the University of North Carolina at Charlotte where he created a cycle-accurate microarchitecture simulator called Casper, which was awarded the best contribution to the OpenSPARC Community Project. His doctoral dissertation used statistical machine learning and Casper to improve power efficiency of simultaneous multi-threading SPARCV9 many core microarchitectures.