This is a Markdown document. If you don't see clickable links in it, please visit it online.
For years I have imagined how a music theory knowledge graph might change how we automate composition. I imagine a unified data set that could answer questions about historical uses of musical techniques (“What is Bach like?”), common or likely emotional associations (“How are car chases scored? What do they play at political rallies?”), lower-dimensional projections of extant music (“What is the typical energy curve for a pop song?”) …
Mining that data from the world, using it to make music, and doing it all in Haskell – it would be a dream come true for me.
My Soundcloud page will demonstrate my music theory and audio-perceptual chops. It includes:
- Music made with Tidal, a Haskell live-coding toolkit. (That’s only 5 lines of code. It takes about 35 seconds to get going.)
- 31-tone equal temperament music made using a monome, Max/MSP, and Java.
- Solo improvised piano
- Improvised guitar + Chapman Stick The improvisation is total; I did not know what chords I would play before playing them. For the last track, when I added guitar, I had to determine by ear what I had previously done on the Stick.
The majority of what I have written using Tidal can be found on Github.
I invented the Reflective Set of Labeled Tuples (RSLT), a data structure that generalizes graphs. In an RSLT relationships can involve any number of members, and those members can themselves be other relationships, as this paper describes. I implemented the RSLT in Haskell, using the Functional Graph Library. You’ll find lots of applicatives and monads in it. I presented it to the Santa Monica Haskell User Group at Brainium.
Semantic Synchrony, the open-source knowledge graph server and Emacs front end, was Joshua Shinavier’s PhD dissertation. I Dockerized it and wrote the documentation for it (including the videos). Also I contribute to the front-end, the public knowledge base, and the collection of git scripts used to coordinate shared graphs.
I am ABD for the PhD in economics at Michigan State University. I completed two papers on econometric theory (a branch of statistics), and had made progress on a third. (My dissertation committee consisted of Gary Solon, Jeff Wooldridge, and Peter Schmidt.)
For the Congressional Budget Office I wrote forecasting software for state-level foreclosure rates. (I reported to Damien Moore, who is happy to vouch for me.)
Precision Health Economics publishes papers in top journals of economics and medicine. There I was a data programmer and statistician. I would integrate data sets from multiple sources (hospitals, drug companies, the Bureau of Labor Statistics, and more) and run regressions and statistical tests on them, to answer questions like, “Which of these drugs saves more lives?” (I reported to Jeff Sullivan, who is happy to vouch for me.)
I implemented a genetic algorithm for evolving electrical circuits to fit arbitrary constraints when I was nineteen, for Lee Spector’s Artificial Intelligence class. I can use and implement feedforward neural networks, support vector machines, and clustering methods.