Vertex AI on Google Cloud

Module 1: Introduction to Vertex AI

• Overview of Vertex AI and its evolution from AI Platform

• Unified experience: AutoML + custom models + pipelines

• Key benefits and components of the ecosystem
  
  

Module 2: Vertex AI Architecture and Services

• Vertex AI Workbench and Notebooks

• Pipelines and orchestration tools

• Feature Store, Model Registry, and Prediction Services

• Integration with BigQuery, GCS, and Dataflow

Module 3: Preparing Data with Vertex AI Feature Store

• Creating and managing feature sets

• Online vs. offline feature storage

• Importing data from BigQuery or CSV

• Feature versioning and reuse

Module 4: Model Training Options

• AutoML: tabular, image, and text training

• Custom model training (via containers or Python scripts)

• Distributed training with managed infrastructure

• Hyperparameter tuning
  
  

Module 5: Model Deployment and Predictions

• Creating and managing endpoints

• Deploying models for online predictions

• Running batch predictions

• Versioning and rollback of models
  
  

Module 6: Building End-to-End ML Pipelines

• Overview of Vertex Pipelines and Kubeflow Pipelines

• Reusable components and pipeline DSL

• CI/CD integration for MLOps

• Pipeline monitoring and debugging

Module 7: Model Monitoring and Explainability

• Monitoring prediction quality and drift

• Alerting and thresholding

• Explainable AI (XAI) with SHAP values

• Auditing models for fairness and transparency

Module 8: Capstone Lab

• Build, train, and deploy a model end-to-end

  • Use BigQuery or GCS dataset

  • Register features in Feature Store

  • Train with AutoML or custom code

  • Deploy and test predictions via endpoint