I’m Thomas Markovich, a research scientist working on machine learning. I’m currently a Senior Staff Research Scientist at Block, and before that I was a Staff Research Scientist at Twitter, where I worked on graph learning and recommendation at the scale of the whole network. Earlier I built knowledge graphs and machine reasoning systems at Forge.AI, and contributed to a probabilistic programming language at Gamalon.

I came to machine learning through physics. My PhD is in Chemical Physics from Harvard, where I worked on methods for correlated harmonic bath models and the treatment of long-range van der Waals interactions, mostly by writing high performance numerical code in C and Fortran. Before that I studied physics and mathematics at the University of Houston. Publications are on Google Scholar, and the longer version of my background is on my resume.

What I write about here

Two bodies of work, separated by a gap of a few years.

The earlier posts are technical: linking Python to C with CFFI, knowledge graph construction, stochastic modeling of life outcomes. They’re still here because they’re still useful, and because I’d rather keep a real record than curate a tidy one.

The recent posts are about what it means to build these systems. Heidegger on technology as a way of revealing the world; Conway’s Law and why automating an org chart automates its politics rather than its work; the question of what genuine inquiry is when the machinery of inquiry can be automated. These are attempts to think clearly about a thing I do professionally, which turns out to be harder than it sounds.

If research notes end up as a third stream, they’ll be filed alongside these rather than just dropped into the date-ordered archive. Sections live on the categories page; everything, in order, is in the archive.

Elsewhere

The RSS feed is the reliable way to follow along. You can also reach me at thomasmarkovich@gmail.com, or on GitHub and LinkedIn.