The new version of NumPy 1.10 contains the new Python @ operator. This is for matrix multiplication and greatly simplifies some code. This also appeals to me as a Math geek because it makes it really easy to write code down based on what you read in a paper. This makes implementing a linear algebra model much simpler.
I had a look at the NumPy chapter in Python for Data Analysis and reimplemented some parts of it using the new operator. I found this a good learning experience