Week 3: ML in the Cloud

This week, we’ll explore how cloud-based storage and machine learning platforms support ML workflows. Students will train and deploy an image classifier using both low-code and code-first tools.

Focus

• Learn about different types of cloud storage.
  • Databricks hive metastore
  • Azure Blob Storage
• Learn how cloud storage fits into ML workflows
• Understand the differences between low-code and high-code approaches
• Get hands-on experience with multiple Azure ML tools
• Know how to visualize data with cloud tools

Hands-On Activities

• Upload an image dataset to Azure Blob Storage
• Build a classifier using:
  • Azure ML Designer (low-code drag-and-drop)
  • Azure ML SDK (Python-based code-first)
  • Databricks AutoML
• Compare pipeline transparency, usability, and deployment steps using the different frameworks.
• Run inference on new images in the cloud

Learning Outcomes

By the end of this week, students will be able to:

• Explain different types of cloud storage (in general, and in Azure).
  • Use appropriate storage formats in an ML workflow.
• Upload and manage image data using Blob Storage
• Build and deploy an ML model using Azure ML Designer
• Use the Azure ML SDK to run inference in Python
• Reflect on the trade-offs between low-code and code-first ML development:
  • Usability
  • Flexibility
  • Cost and compute usage
• Know how work with features and metrics with data visualization tools

Resources

• Azure Blob Storage Overview
• Azure ML Designer
• Azure ML SDK Documentation
• Quickstart: Train and Deploy ML Model
• Databricks AutoML

Instructor Notes

We will need to decide whether students will build their own pipelines or collaborate in small groups.