How to Deploy Faster Machine Learning Workloads on AKS Clusters

Learn how to build secure, scalable, and accelerated Kubeflow* pipelines on a Microsoft Azure* Kubernetes* Service (AKS) cluster.

Intel® Cloud Optimization Modules are cloud-native, open source reference architectures designed to help developers build and deploy AI solutions on leading cloud providers.

This session focuses on a module for Azure—Kubeflow Pipeline with XGBoost—and walks you through the process of building confidential Kubeflow pipelines for machine learning using Intel-optimized software and security technology on an AKS cluster.

This module shows you how to:

• Configure Azure cloud services (including an Azure resource group and Azure container registry) and AKS cluster with a confidential computing node pool.
  
• Install the Kubeflow software layer on an AKS cluster using a secure transport layer security (TLS) protocol.
• Implement a full, end-to-end machine learning pipeline using an XGBoost model plus Intel® AI software optimizations to predict credit risk, from data preprocessing to model inference.
• Take advantage of the security of confidential computing nodes on an AKS cluster using Intel® Software Guard Extensions (Intel® SGX) and use Intel® Turbo Boost Max Technology 3.0 to reach 3.5 GHz for AI acceleration.

The presentation includes a hands-on demo showing how to set up and run the full Kubeflow pipeline.

Skill level: Intermediate

• Get the following as stand-alone tools or as part of the AI Frameworks and Tools:
  • Intel® Optimization for XGBoost*
  • Intel® Extension for Scikit-learn*
• Get the stand-alone version of Intel® oneAPI Data Analytics Library (oneDAL) or as part of the Intel® oneAPI Base Toolkit. 

Kubeflow Pipelines with XGBoost on Azure

See All Code Samples