An introduction to what Docker is and why and how to use it for Kaggle.

Recurrent neural network gradients by hand.

A post-mortem statistical simulation of the 2016 Colombian plebiscite.

A Python library for some canonical neural networks, with a self-righteous emphasis on readability.

Performing gradient descent on a single neuron.

A web app for Spotify music recommendation using Implicit Matrix Factorization and "country arithmetic."

Implementing custom loss functions in Python using autograd.

A Scala-based web app that gives travel recommendations via Twitter data and the Jaccard similarity.

Auto-generating travel blog posts via Markovian processes.

Predicting the difficulty of the road ahead: a bike ride through Southeast Asia.