Data science has a strong focus on the Python and R programming languages. But there are plenty of other programming languages out there, that are used for a variety of applications (irrespective of how much of a Python-maximalist people can be these days). If ML is not the main focus of an application, forcing a programming language for a small feature in a larger whole makes no sense. Sure, most ML can be packed as an API but it still needs a separate server and expertise in the secondary programming language. Instead, each programming language is slowly getting they fair share of ML toolkits.
This idea hinges on two projects I care about: a CRM for friends, Monica and the RubixML project. Both projects are written in PHP, one of the languages with larger install bases out there.
Many years ago, I did some ML consulting for a now defunct company in the friendship space. It turns out that with enough data, patterns emerge from friendships. Data science can help people maintain healthy relationships with friends and family. Or at least that is the hope.
As a friend CRM, Monica allows its users to log calls, activities and keep a running diary of friend-related information and events. Adding some small features to Monica using the machine learning algorithms contained in Rubix can showcase the value behind keeping the information up-to-date.
For example, some of Monica philosophy involves more self-reflection about how interacting with other people make us feel. This can be further explored by building a decision tree on the different activities recorded and showing the user a visualization of the decision tree.