An overview of the ML model development life cycle

You may be familiar with the concept of the software development life cycle (SDLC), which is taught in computer science classes in schools all over the world. The SDLC concept began to be formulated in the 1960s and early 1970s, and by now it is a well-established and well-understood process that is used in various formats by pretty much every company that develops software. Without formalized processes for people to follow when developing software, it would be difficult for companies to efficiently produce high-quality software and the software development industry would be quite chaotic. In fact, that’s how the software development industry was in its early years, and that’s how the machine learning industry currently is for most companies. Only in the past couple of years has the industry started to establish some structure around how companies should develop ML models and their related applications.

In this section...