How to build a DIY deep learning framework in NumPy

As a general rule of thumb for learning, you will only understand something in detail once you are able to build it on your own. This is particularly true for technical subjects like machine learning and neural networks. During your first encounter with deep learning, you can take a few large steps by learning and mastering a high level framework such as TensorFlow or PyTorch. However, if you want to go really deep, you need to go back to square one: you need to build your own neural network and optimizer from scratch! This is a common rite of passage for deep learning engineers and researchers.