`reticulate` is a brilliant package for running Python within R. I guess it's good as long as it's local machine. I've not tried productionizing a code that has both R+Py. But it can work within Shiny too, helping bring Python's Datascience stack (esp. scikit-learn) within R.