Saurabh Kulkarni is currently the Head of Engineering in North America at Graphcore. In his career spanning 20 years, he has held leadership positions at Intel, Microsoft and Oracle prior to his current role at Graphcore. His roles have spanned Computer Architecture, Server Platform Architecture, Cloud infrastructure design, Virtual Machine sizing for Microsoft Azure, HPC infrastructure in the cloud and more recently hardware accelerators and scale-out infrastructure for AI compute in the cloud.
After graduating from the University of Minnesota, Saurabh started his career at Intel Corporation, where he initially worked on several generations of X86 CPUs for client and server market segments. Towards the later part of his time at Intel, he was a platform architect in the data center group driving customer focused innovation in the enterprise platform and firmware security areas.
In 2014, Saurabh joined Microsoft in the Azure Compute team focusing on the IaaS platform and infrastructure, with particular emphasis on scale-out architecture. He drove the server platform architecture and Virtual Machine (VM) sizing strategy for various Azure VM families across general purpose and accelerated computing use cases including HPC/AI workloads. In this capacity, one of his primary areas of focus was efficient architectures for disaggregated resource management for general purpose and accelerated compute.
Saurabh briefly worked at Oracle in the Oracle Cloud Infrastructure team working on cloud native services before joining Graphcore, where he currently is the Head of Engineering for North America. In his current role, he works closely with customers to understand their AI workload needs for training and inference with the objective to devise efficient, end-to-end solutions that can be either deployed on-prem, on public clouds or in a hybrid cloud setting. He is also involved in the architectural definition of the next generation of Graphcore IPUs and systems being designed to fulfil the demanding needs of future AI workloads at scale.
