Episode

FastTrack for Azure Season 2 Ep04: Azure ML Operations

with Andres Padilla, Meer Alam, Kris Bock

This live session will introduce the concepts of MLOps, discuss how organizations can adopt MLOps practices, and develop a machine learning lifecycle from a technical perspective.

Learning objectives

  • Operationalization of an end-to-end ML solution through Azure ML and DevOps practices

Chapters

  • 00:00 - Welcome
  • 02:44 - Introduction
  • 03:23 - Learning Objectives
  • 07:27 - Over of Machine Learning Operations (MLOps)
  • 09:52 - Walkthrough the Mindmap of MLOps
  • 16:40 - Data Preparation for your ML Model
  • 20:55 - Azure ML Computes for your Model training
  • 23:31 - Model Training Environment
  • 25:21 - Azure Databricks has similar components like Data Assets, Compute, Environment like we have in Azure ML
  • 29:07 - Authoring your models - Azure ML Studio Designer, Automated ML and Notebooks
  • 32:36 - Azure ML Jobs
  • 35:41 - Use Components to create Azure ML Pipeline
  • 41:30 - Deploying your models to an Endpoint - Model Development Cycle Completed in Azure ML Studio
  • 44:40 - How Does Registry help in MLOps?
  • 47:41 - Leverage Azure ML Python SDK V2 for your Model Training
  • 01:06:22 - Demo - ML pipelines with Python SDK V2
  • 01:27:00 - DevOps for ML Solution - Walkthrough with Azure DevOps and GitHub Action to automate end-to-end operationalization of ML Solutions in Azure.
  • 01:28:04 - Summary and conclusion

Connect

Intermediate
AI Engineer
Data Scientist
Data Analyst
Azure Machine Learning