The MLOps Workbook: A Guided Online Course for Getting Started with MLOps

This course will help you and your team understand the concepts and best practices needed to scale up your Machine Learning Operations (MLOps) in your machine learning projects. It provides ML project team members with an extensive overview of the challenges and decisions encountered in building professional ML systems. Rather than focusing on evolving tools, the course emphasizes concepts and frameworks that help to share a common understanding of MLOps within ML teams. The course is structured around the ML Lifecycle, a key perspective on MLOps, from planning a machine learning project to implementing feedback loops after your project is deployed. Among other things, you will learn about how to plan an ML project, how to apply the appliedAI Project Management Framework to your ML projects, how various accountabilities should be involved during the different project phases and how the ML Principles function as the foundation of MLOps. The course comes with a video series and a workbook that you can keep to easily access the course content. You work with your own copy of the workbook in the form of a PDF file, digital whiteboard, or physical copy and work alongside the video series.

🎯 Upon completion of the course, we expect you to be able to explain the fundamental principles and frameworks underlying MLOps. You should be able to discern the differences between professional and unprofessional MLOps workflows, thereby empowering you to identify and implement tangible improvements within your own MLOps processes.

📌 In particular, upon completion you should be able to:

• Explain the four perspectives on MLOps, namely the ML Lifecycle, the ML Accountabilities, the ML Principles, and the appliedAI Project Management Framework
• Explain the key considerations and best practices necessary for effective project planning in an ML project
• Name and describe specific improvements applicable to each stage of the ML Lifecycle
• Explain the involvement and responsibilities of the various ML Accountabilities throughout the ML Lifecycle

You can work with the workbook only or you can use the workbook in parallel with the video course.

The latest version of the workbook is not compatible with the video course. We are constantly developing the content.

Overview of the four perspectives on MLOps

This module will cover the four perspectives on MLOps, including the ML Lifecycle, the ML Accountabilities, the ML Principles, and the appliedAI Project Management Framework. You will learn how these concepts relate to one another to set the stage for the subsequent modules.

The Scoping Stage of the ML Lifecycle

This module will delve deep into project planning, the first stage of the ML lifecycle. You will learn how to break down an AI project into phases in order to mitigate uncertainties. In addition, you will learn how to plan AI projects from conception to deployment by conducting a project planning workshop.

The Data Engineering Stage of the ML Lifecycle

This module will showcase various methods to improve the data engineering in your organization by following best practices during data ingestion, data preparation, and data management. Among others, you will learn about techniques such as ETL/ELT, data versioning, automatic quality checks, tracking data lineage, and data catalogs.

The Modeling Stage of the ML Lifecycle

This module will showcase various methods to improve model training and model management. You will learn what an ML experiment is and explore techniques such as experiment tracking, model versioning and model registries. Also, you will learn how Data Scientists and ML Engineers can effectively collaborate during the Modeling Stage.

The Deployment Stage of the ML Lifecycle

This module will showcase various methods to improve the deployment management, monitoring and maintenance of AI models. You will learn about different model-serving patterns and deployment strategies. Also, you will learn how model performance is affected by data drift, covariate shift, label shift, and concept drift.

Feedback Loops & ML Orchestration

This module will discuss the various feedback loops throughout the ML lifecycle and discuss ways to orchestrate its execution. You will learn what it means to orchestrate an ML workflow, what types of triggers exist that warrant a model retraining, and how workflow orchestration tools support collaboration within ML teams.

Access the course here and get started with MLOps.

Our content, available under the Creative Commons Attribution 4.0 International License (CC BY 4.0), can be freely shared, adapted, and used for commercial purposes with proper attribution to "AppliedAI Institute for Europe gGmbH."