Machine Learning on Google Cloud (Vertex AI) - Hands on!

Create a free GCP account at cloud.google.com/free, claim $300 credit for 90 days, and sign in with a personal Gmail for an individual project.

Explore Google Cloud compute resources, including virtual machines, containerization with Kubernetes Engine, Cloud Functions, and App Engine, to run machine learning workloads and manage costs.

Master identity and access management to grant least-privilege access to Google Cloud resources by defining roles and permissions for users, groups, and service accounts.

Explore identity and access management on GCP by granting and refining bucket and project permissions, testing read-only and full access, updating policies, and creating custom IAM rules for storage.

Explore identity and access management basics on Google Cloud: grant or deny access to projects and storage with fine-grained controls, using member identities, custom rules, Google Groups, and service accounts.

Leverage Dataflow for real-time transformation, Dataprep for cleaning and preparing data, and BigQuery for serverless warehousing and analytics.

Create a BigQuery dataset and table from a local file for a baby names dataset in a machine learning project, run queries, and visualize results in Data Studio.

Learn to build a dataprep workflow in Vertex AI by importing data, applying recipes, handling missing values, filtering and joining datasets, and saving output to cloud storage with scheduling.

Explore Google Cloud's AI services, from pre-trained models to AutoML and platform options, and learn how Vertex AI enables easy deployment and production of ML models.

Discover how AutoML on Vertex AI enables training, evaluating, and deploying custom models with minimal data science effort, using vision, language, and audio services.

Evaluate the trained model using precision, recall, and confusion matrix; deploy the model and generate predictions via Python or cloud storage, using the model ID.

Configure hyperparameters with a hyperparameter configuration file, run trials to maximize accuracy, and deploy the trained model to a regional endpoint on Vertex AI, including model versioning and testing.

Enable cloud build, create a Docker image from Dockerfile, install TensorFlow via a requirements file, push to Google Cloud container registry, and run on Compute Engine with AI Platform notebooks.

Train a custom model on Vertex AI with a docker image using the Grab dataset to predict crop age from diameter, height, and weight, then deploy for predictions.

Deploy a Vertex AI custom model to an endpoint, set 100% traffic, select a standard eight machine type, and reuse Python code for online predictions after normalization.

Build and deploy a Vertex AI AutoML image classification pipeline in Google Cloud. Create datasets, train models, evaluate with thresholds, and deploy endpoints within the pipeline.