My favorite machine learning development system, by far, is PyTorch, not only because of its wide adoption in recent years, but because it was the language that really helped me understand what truly goes on within machine learning systems at the beginning of my career, and is still practical to use in the advanced systems I create today. I highly recommend anyone looking into computer vision based on deep learning to take a look at PyTorch. For non deep-learning systems (as well as algorithms to support such systems), I also recommend the NumPy-derivative suites of libraries, including NumPy, SciPy, SciKit-Learn and OpenCV. OpenCV specifically defines a large number of classic computer vision algorithms that can be used as components in larger systems, and I use OpenCV functions in my research to this day.
One of the less obvious (and potentially controversial) advantages of attending schools with large financial resources and academic relevance is the ability to deal with large amounts of gaslighting and abuse and be relatively unfazed by it. Ivy+ institutions are brutal. Having the resilience to not only advocate for yourself and maintain high performance while balancing a multitude of projects is a skill I've seen in almost every student who has made it through an ivy-caliber program on their own merit, and as a result, most of these students can be highly successful navigating both the technical and social aspects of any job they take after graduation.