About
I'm Steven Fazzio. I'm a data scientist, but the through-line of my work is closer to "storyteller who codes." These days that mostly means making maps of large text collections: take a corpus (every game on Steam, five years of research papers, a video store's rental catalog), embed it, lay it out in two dimensions, and label the regions so you can actually explore the thing. If that sounds abstract, the projects page is the fastest way to see what I mean.
I've worked at tech startups on marketing, data science, and engineering teams. I wrote my first (very small) language model about twenty years ago and have been hooked on computing with natural language ever since. My current interests sit at the intersection of NLP, embeddings, and data visualization, with a growing conviction that interactive visualization is a narrative medium: a good map isn't a chart, it's a place to wander around in.
This site is the canonical home for my work. Every project gets a write-up here, the interactive artifacts live at subpaths of this domain, and everything is available by RSS.
I'm available for consulting through Fazzio Consulting. If you have a large pile of text and questions about what's inside it, get in touch.
You can also find me on GitHub.