Detailed Course Outline
Module 01 Architecting low-code AI solutions
Topics
- Ira needs to understand customer segments using BigQuery and a clustering model.
- Sasha needs to predict customer value using AutoML Cymbal Retail’s customer dataset.
- Taylor needs to build a conversational AI assistant for customers using Vertex AI Agent Builder and retrieval-augmented generation (RAG)
- Diagnostic questions
- Review and study planning
Objectives
- Identify your level of knowledge in developing and implementing BigQuery ML and AutoML machine learning solutions.
- Determine the skills needed to select appropriate ML APIs, prepare data effectively, and build custom models using AutoML.
Activities
- Lecture
- Diagnostic questions
- Quiz
Module 02 Collaborating within and across teams to manage data and models
Topics
- Use Google Cloud's products and Cymbal Retail's rich data to design a model to predict which high-value customers are likely to stop purchasing (also known as customer churn).
- Answer diagnostic questions.
- Review the information and plan your study.
Objectives
- Identify your level of knowledge in exploring, preprocessing, and managing organization-wide data.
- Identify your level of knowledge in addressing privacy implications and leveraging tools like Vertex AI Feature Store.
- Determine the skills needed to prototype models using Jupyter notebooks on Google Cloud.
- Determine the skills needed to select appropriate backends, implement security best practices, and integrate with code repositories.
Activities
- Lecture
- Diagnostic questions
- Quiz
Module 03 Scaling prototypes into ML models
Topics
- Use Google Cloud's products and Cymbal Retail's rich data to build and scale customer churn prototype into a production-ready model
- Answer diagnostic questions.
- Review the information and plan your study.
Objectives
- Identify your level of knowledge in scaling ML prototypes into production-ready models
- Identify your level of knowledge in selecting appropriate ML frameworks, model architectures, and modeling techniques based on interpretability requirements.
- Determine the skills needed to train models effectively, including organizing and ingesting training data on Google Cloud.
- Determine the skill needed to utilize distributed training techniques, perform hyperparameter tuning, and troubleshoot training failures.
Activities
- Lecture
- Diagnostic questions
- Quiz
Module 04 Serving ML models
Topics
- Use Google Cloud's products and Cymbal Retail's rich data to deploy a customer churn model and use it in production for inference.
- Answer diagnostic questions.
- Review the information and plan your study.
Objectives
- Identify the level of knowledge needed to effectively serve models in production.
- Identify the level of knowledge needed to select between batch and online inference, utilize various serving frameworks, organize a model registry, and conduct A/B testing for model optimization.
- Determine the skills needed to scale online model serving, including leveraging Vertex AI Feature Store.
- Determine the skills needed to manage public and private endpoints, choose appropriate hardware, optimize serving backends for throughput, and fine-tune models for optimal performance in production.
Activities
- Lecture
- Diagnostic questions
- Quiz
Module 05 Automating and orchestrating ML pipelines
Topics
- Use Google Cloud’s products to orchestrate the entire machine learning pipeline for seamless execution and continuous improvement with customer churn.
- Answer diagnostic questions.
- Review the information and plan your study.
Objectives
- Identify the level of knowledge needed to develop and maintain end-to-end ML pipelines.
- Identify the level of knowledge needed to validate data and model, consistent preprocessing, hosting options, component identification, parameterization, triggering mechanisms, compute needs, orchestration strategies.
- Determine the skills needed to automate model retraining, including establishing retraining policies.
- Determine the skills needed to implement CI/CD model deployment, and track and audit metadata (model artifacts, versions, data lineage).
Activities
- Lecture
- Diagnostic questions
- Quiz
Module 06 Monitoring ML Solutions
Topics
- Use Google Cloud’s products to ensure the customer churn model remains robust, reliable, and aligned with Google’s Responsible AI principles.
- Answer diagnostic questions.
- Review the information and plan your study.
Objectives
- Identify the level of knowledge needed to assess and mitigate risks in ML solutions.
- Identify the level of knowledge needed to build secure ML systems, align with responsible AI practices, evaluate solution readiness, and utilize model explainability on Vertex AI.
- Determine the skills needed to monitor, test, and troubleshoot ML solutions.
- Determine the skills needed to establish continuous evaluation metrics, monitor for training-serving skew and feature drift, compare model performance against baselines, and investigate common training and serving errors.
Activities
- Lecture
- Diagnostic questions
- Quiz
Module 07 Your next steps
Topics
- A sample study plan for the exam
- How to register for the exam
Objectives
- Review a sample study plan for the exam
- Learn how to register for the exam
Activities
- Create your study plan for the exam
- Identify a date to take the exam based upon your plan
- Register for the exam