It all started with a simple classification problem: Given a document, identify what company it was about. Here are 10⁶ documents we know how to classify, and here are 10⁵ we don't. A standard undergrad homework exercise in machine learning, they said.

Little did I know I was headed down the rabbit hole…

A year later, I feel like I'm just beginning to get my head around all this and I'm going to try to get it down in a form that makes some sense to me.