https://blog.concurlabs.com/10-differences-between-ml-and-software-development-part-1-3e21a83094d1

About 5 years ago, I wrote a post on a list of 50+ machine learning APIs. It got a fair amount of likes and shares at that time*. As a software developer, APIs provided a safety net from where I can interact with ML without having to understand how it works.

Nowadays I can’t go between my morning coffee and lunch without at least hearing two people in a cafe talk about AI, neural networks, epoch, TensorFlow, training your model, ML-powered , etc. At work I would discreetly Google terms here and there, just to keep up with the growing list of ML terminology going over my head. Eventually I decided I’d put in the time to learn.