An overview of how the NumPy family of libraries help make Python more performant for scientific computing, without losing the benefits Python brings.
Brian gives a deep explanation of how Monte Carlo Tree Search (a key technique for game AIs) works. He implemented the AlphaGo paper from scratch, so he should know!
An explanation of metrics for comparing statistical classifiers, and why the most common one, AUROC, is so commonly used.
A minimal Go AI modeled on AlphaGo. It became the basis of Minigo (https://github.com/tensorflow/minigo), which is maintained by the Tensorflow team.