I've spent more than a decade somewhere between mathematics and a terminal: a PhD scientist who never quite stopped being an engineer. The work runs from numerical solvers on thousand-node HPC clusters to RESTful APIs, database schemas, and the occasional stubborn UI pixel.

Along the way I've written code in Julia, Python, C, C++, Fortran, JavaScript, TypeScript, Java, Scala, Groovy, and a few others I poked at just to see how they felt. Ask me which language is best and I'll ask whether a spoon beats a fork: tools are tools, and the good ones are all great once you know what you're reaching for.

I founded and maintain ReactiveBayes (25+ open-source Julia packages), built a database that ended up in Nature and has quietly served researchers for 10+ years, and spent two years as a startup CTO learning that sales is harder than message passing. I was reinstalling Linux and patching kernels long before either got easy, and ran Docker back when it still raised eyebrows.

If all of that taught me one thing, it's that software is meant to last: the throwaway script becomes load-bearing, and someone is still running it years later. An LLM can spin up a calculator in seconds; I'd rather build the one that still compiles, deploys, and passes its tests a thousand years from now. Which is the long way of saying I write tests first and mean it. What follows is the work that survived all of that.