R is not meant for GIS - it can handle some geo data in tables, but it is really no match for PG or ArcPy. R is much better for statistical analyses and one of my colleagues does most of his work in Excel or R because of it. He’s the only non-geo data scientist, though, and he doesn’t use R’s geo abilities because it’s not powerful enough.

PG can do a lot, especially if you’re clever at combining the rather-basic-but-incredibly-fast functions it offers. (And then there’s always QGIS.) ArcPy has a larger suite of functions, but only within the realm of things Esri wants you to do, so if you’re out there doing anything groundbreaking, you run into barriers more easily.

By the way - there’s an ArcGIS For Home license which gives you an Advanced ArcGIS license (including Pro) for $100/yr if you pinky promise to never use it commercially. I used it for my own experiments.