Operationalize AI

How to get the buzz out of AI!.In this course you will get a practical introduction into how you can go from a machine learning model to production. This course will cover the whole flow from exploratory data analysis (EDA), feature engineering, building and testing your machine learning model to deploying the model in a container and expose its functionality as an api. The focus of this course is on the latter part but we will cover the first part since this is part of the overall process.

The course will cover different types of production flows for the machine learning model: live, batch, precompiled and cached. We will cover different types of deployment: to cloud, on-premise and directly onto device

An important aspect when you put machine learning into production is how you will be able to do monitoring and testing. In testing it is important to both test the code but also the data itself. In monitoring we will discuss how to model output and how to monitor for data drift.

Course Objectives:

In this course, you will learn to:

• Different production flow for machine learning
• Limitations due to deployment type
• Deployment of machine learning models through containers  
• Making APIs for machine learning models
• How to modularize feature engineering, building and testing of machine learning models
• How to test machine learning models
• How to monitor live model performance

Audience:

• Developers
• Architects
• Testers / QA
• Data Scientist

Prerequisites:

Participants need to know python as a programming language. This course is both for those who have experience with machine learning but not with deployment and for those with experience in deployment but not in machine learning.

Course outline:

• Introduction
• Overview of machine learning pipelines
• Overview of deployment options and their effect on model selection
• Modular feature engineering and training
• Testing
• Deployment through containers
• API
• Monitoring

 

Other relevant courses