Teaching is the best way to learn. While I was at Square, I led SquareU tutorials on Python for machine learning / neural networks every quarter or so. If you're interested in any of the tutorials below, please fill in the form here.
Intro to Python for machine learning
This tutorial starts with data manipulation using pandas to clean up the data. We'll then use scikit-learn to make predictions. By the end of the 1-hour session, we would have worked on the Kaggle Titanic dataset from start to finish, through a number of iterations in increasing order of sophistication. The first version of this tutorial was delivered at PyCon UK 2014 (reviews here and here, Korean translation here). The most recent session was September 2021.
Intro to Python for neural networks
This tutorial is designed to get the audience training neural networks by the end of a 1-hour session. In particular, it covers areas where neural networks really shine - CNNs and RNNs. These techniques are applied on the Kaggle Titanic, MNIST and Rotten Tomatoes datasets. The most recent session was October 2021.
Applying machine learning to payments
This is a case study on applying machine learning to flag suspicious payments. It assumes no prior knowledge of payments or machine learning, and is designed for a general audience. The was first presented at Blackrock, and later at Metis.