Learn to productionize machine learning pipelines using Databricks and Azure ML in this 49-minute tutorial. Discover a reproducible framework for quickly launching data science projects that enables easy production-ready app deployment. Explore sample code, templates, and recommended project organization structures while gaining insights into deploying machine learning pipelines. Develop pipelines for continuous integration and deployment within Azure Machine Learning using Azure Databricks. Execute Apache Spark jobs with Databricks Connect and integrate source code with Azure DevOps for version control. Gain hands-on experience building deep learning models for image classification using TensorFlow and PyTorch. Address ML lifecycle challenges by implementing MLflow to track model parameters, results, package code for reproducibility, and deploy models. Prerequisites include an Azure account, configured workspaces, Databricks trial, Docker installation, and basic knowledge of Python, Spark, and deep learning concepts.
Overview
Syllabus
Introduction
Local Environment Configuration
Virtual Environment Configuration
Poetry
Bundled Fault
Project Layout
Business Logic Code
Azure ML Workspace
Code Block
Authentication
Creating the Cluster
Running the Code
Attaching Databricks to Compute Target
Mounting Blob Store to Databricks
Setting up Databricks Connect
Running Databricks Connect
Data Set and Preprocessing
Tensorflow Training
Code Overview
ML Flow Source Code
Getting the Best Model
Registering the Model
Testing the Model
Azure ML Pipelines
Pipeline Infrastructure Code
Continuous Integration Phase
Taught by
Databricks
