Linear algebra is the most widely used branch of mathematics in machine learning. However, linear algebra itself can be rather dry, if not viewed geometrically. So it is very important to look at linear algebra from a geometric point of view. Please study this geometric review of linear algebra and slides.
Python is super handy for symbolic programming and data processing, but lacks support for numerical computation and linear algebra (such as vector/matrix/tensor operations). On the other hand, Matlab is super good at numerical computation, but is not good for symbolic programming or general data processing. Numpy bridges the two, providing Matlab-style syntax and numerical capabilities for Python. Nowadays, Python + numpy has become the default way of doing ML, and there are many other ML/DL toolkits based on numpy, such as scikit-learn (sklearn), scipy, matplotlib, tensorflow, and pytorch.
Please study this official numpy tutorial and watch our videos.