Azure ML E2E Workshop

Target Audience

Anyone who wants a comprehensive E2E understanding of Azure ML.

Key Goals

1. Understand the product E2E
2. Ensure that the work we are doing in Mn will address key gaps in usability / E2E user experience
3. Open bugs, fix docs & ensure commitments from product area leads on whether / how those bugs will be fixed

Agenda

Workspace Concepts: infra setup, ARM, workspace setup, computes, datastores, setup**

1. Set up your workspace and compute
2. Register a dataset
3. Run AutoML from the UI
4. Compute Instance - Clone Git Repo

Datasets, Model Training (AML, HyperDrive and AutoML), Model Inference**

Notebooks to run and research:

AML training, HyperDrive and Interpretability:

• Notebook for plain vanilla Scikit-Learn model training in AML local compute (AML VM)
• Notebook for Scikit-Learn model training in AML remote compute and HyperDrive
• Notebook for Model Interpretability in AML

Automated ML:

• Notebook for AutoML local compute
• Notebook for AutoML remote compute

Pipelines & Batch Inference

• Use a model for batch inference

MLOps (model management, deployment, inference, automation)

Tutorials for MLOps

• Deploy a model
• Automate training & deployment
• Create automation workflow with EventGrid

Enterprise Readiness

• Azure Monitor https://docs.microsoft.com/en-us/azure/machine-learning/service/monitor-azure-machine-learning
• RBAC https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-enterprise-security
• https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-assign-roles
• Limits service https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-manage-quotas
• VNET https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-enable-virtual-network