I have a script that retrieve photos from an attached camera and then uploads the images to an AWS GPU instance and train an image recognition model there. This is all for controlling my model trains :-)