Kaarthik Sivashanmugam

About Kaarthik Sivashanmugam

Kaarthik works in the Tanzu group at VMware enabling organizations to package, run and manage cloud-native applications efficiently on public and private clouds using a unified application platform. In his current role, he is supporting a global engineering organization for a product that is loved by customers and brings $500 million in annual revenue. He has expertise in building data platforms, machine learning platforms/frameworks/workloads, and cloud services. Previously, Kaarthik worked at the Azure AI group at Microsoft enabling ML practitioners to optimize and scale model training. He led various engineering efforts and supported teams of engineers and data/applied scientists. His work at Microsoft include:* Acceleration of PyTorch implementations of large Transformer model training using ONNX Runtime * Modernized Bing Core Search Relevance Ranking ML pipeline using Azure ML * Azure Machine Learning service * Distributed deep learning platform leveraging GPU cloud * Platform for compliant ML to train models using eyes-off datasets * Mobius.NET binding to Apache Spark and implementation of Spark Streaming workloads for processing Bing logs * Near real-time analytics platform (Kafka, Apache Storm and Elasticsearch) and implementation of data processing pipelines for Bing Ads * Data Quality Service & SQL Server Integration Services Kaarthik is a frequent speaker at premier industry conferences. His recent conference talks are:* Accelerated Training of Transformer Models, Data+AI Summit EU 2020 * Infrastructure for Deep Learning in Apache Spark, Spark + AI Summit, CA 2019 * Accelerated Data Science Pipeline with RAPIDS on Azure, GPU Technology Conference, CA 2019 * Distributed DNN Training: Infrastructure, Challenges, and Lessons Learned, Artificial Intelligence Conference, NY 2018 * Spark at Scale in Bing: Use cases and lessons learned, Strata Hadoop World, CA 2017 * Mobius: C# Language Binding for Spark, Spark and AI Summit, CA 2016 Blogs on his work:* https://techcommunity.microsoft.com/t5/azure-ai/onnx-runtime-training-technical-deep-dive/ba-p/13•••10 * https://techcommunity.microsoft.com/t5/Azure-AI/Bing-accelerates-model-training-with-LightGBM-and-Azure-Machine/ba-p/10•••99 * https://techcommunity.microsoft.com/t5/Azure-AI/Accelerating-Distributed-Training-in-Azure-Machine-Learning/ba-p/10•••50 * https://azure.microsoft.com/en-us/blog/microsoft-makes-it-easier-to-build-popular-language-representation-model-bert-at-large-scale/* https://azure.microsoft.com/en-us/blog/azure-machine-learning-service-now-supports-nvidia-s-rapids/

Experience

1. Director of Engineering

   Vmware Tanzu

   Feb 2021 — Present

   https://cloud.vmware.com/tanzu https://github.com/vmware-tanzu