Python for DevOps

Chapter 14. MLOps and Machine learning Engineering

One of the hottest job titles in 2020 is machine learning engineer. Other hot job titles include data engineer, data scientist, and machine learning scientist. While you can be a DevOps specialist, DevOps is a behavior, and the principles of DevOps can be applied to any software project, including machine learning. Let’s look at the some core DevOps best practices: Continuous Integration, Continuous Delivery, Microservices, Infrastructure as Code, Monitoring and Logging, and Communication and Collaboration. Which of these doesn’t apply to machine learning?

The more complex a software engineering project is, and machine learning is complex, the more you need DevOps principles. Is there a better example of a Microservice than an API that does machine learning prediction? In this chapter, let’s dive into the nitty-gritty of doing machine learning in a professional and repeatable way using a DevOps mindset.

What Is Machine Learning?

Machine learning is a method of using algorithms to automatically learn from data. There are four main types: supervised, semi-supervised, unsupervised, and reinforcement.

Supervised Machine Learning

In supervised machine learning, the correct answers are already known and labeled. For example, if you wanted to predict height from weight, you could collect examples of people’s heights and weights. The height would be the target and the weight would be the feature.

Let’s walk through what an example of ...