Explore the integration of MLflow and Azure Machine Learning for efficient ML lifecycle management in this 44-minute conference talk by Nishant Thacker from Databricks. Discover how this powerful combination addresses bottlenecks in ML projects and enhances MLOps capabilities on Azure. Learn about versioning, run history maintenance, production pipeline automation, cloud and edge deployment, and CI/CD pipelines. Gain insights into the ML lifecycle management process, MLflow components, and the MLflow model lifecycle. Follow along with a comprehensive demo that covers installing MLflow, setting up the API, importing SDKs, and working with Azure Machine Learning Workspace. Dive into practical aspects such as creating sample applications, utilizing MLflow API for experiments and tracking, building images, testing environments, registering models, and managing metrics and deployments. Understand the importance of governance and usage quotas in ML projects. By the end of this talk, grasp how MLflow and Azure Machine Learning synergize to streamline and strengthen the entire machine learning lifecycle.
MLflow and Azure Machine Learning - The Power Couple for ML Lifecycle Management
Databricks via YouTube
Overview
Syllabus
Intro
Why ML Lifecycle Management
Machine Learning Process
ML Flow Components
ML Flow Model Lifecycle
ML Ops
Demo
Install MLflow
Set MLflow API
Import SDKs
Azure Machine Learning Workspace
MLflow Tracking URI
Create Sample Application
Create Model Script
MLflow API
MLflow Experiments
MLflow Tracking Server
What Next
Build Image
Test Environment
Register Model
Metrics
Deployments
Governance
Usage Quota
Summary
Taught by
Databricks
