About me

I am Paul Metcalfe, am an applied mathematician by training, and have been doing semi-useful things in drug development for the past 14 years, for most of that leading a team of statistical / computational experts. Before my pharma phase I was briefly the world expert in silent underwater breakfasts¹.

JAX: is the nicest way in the world to write the kind of linear algebra that’s used for things like deep learning (or, more usefully, for bayesian things).
blackjax: is my current favourite way to do Bayes Stuff.
rust: current favourite programming language (I’ve been through C++, python, common lisp, haskell, and various others all the way back to BBC BASIC) but I’m now on rust.
Apache Arrow: if you do do data and do not do Arrow you are behind the times.
horseshoe prior: and other ways to sanely do inference in high dimensions.
TRIPOD: what you actually have to think about if you want to make statistical prediction models actually work².
dog training: I have a Doberman and a working-line spaniel, and do a bunch of (mostly) positive training with them. I really want a Malinois, but lack the necessary 48 hours in the day.