Ever want to try implementing a complex software system but thinking you don't have enough compute resources to do so or that you simply cannot afford them? Well, there is a way around this and it's run by none other than one of the major leaders in computing, Google.
Google Cloud Platform offers $300 in compute credits to any new user over the course of a year. In this time, you can try doing computationally expensive tasks (machine learning, stochastic modeling) and pick up working in a Linux environment from scratch. The steps to sign up and get started with a powerful setup are fairly simple and most credit redemptions work on non-student Google accounts. These compute resources include GPUs which have acquisition costs of over $1000. You get to write code for this hardware without having to purchase it!
If you are interested in getting on GCP, feel free to click any of the tutorial links below. If you specifically have an interest in machine learning as it applies to computer vision, feel free to follow the tutorial from CS 231N (Convolutional Neural Networks) as it will provide you with the necessary tools to get started on your own computing cluster!
http://cs231n.github.io/gce-tutorial/
https://cloud.google.com/getting-started/
https://cloud.google.com/compute/docs/tutorials