About

About

Hi, I’m Stephen Bonner, a machine learning researcher based in Cambridge, UK, currently working in pre-clinical drug discovery at AstraZeneca. (Views expressed here are my own and don’t represent my employer.) I hold a PhD in machine learning from Durham University, and spent several years across postdoc and research roles, including a joint postdoc between AstraZeneca and Mila, before joining AstraZeneca full-time.

I’m particularly interested in applying LLMs to scientific problems, GPU performance, and Linux-based systems. This blog is where I document experiments, deep-dives, and lessons learned, especially when things don’t go quite as planned.

Outside of work, I enjoy cycling, skiing, photography, and tinkering with hardware and home lab setups. These topics may occasionally show up here too.

Feel free to get in touch on LinkedIn.

Current Focus

A few things I’m currently digging into:

  • LLMs for scientific workflows

  • GPU performance and inference optimization

  • Reproducible Linux-first ML tooling

Tools Stack

Some of the tools and technologies I enjoy using (in no particular order and by no means an exhaustive list) include:

Operating Systems

  • Arch Linux: My primary OS for development, experimentation and compute. Checkout my dotfiles for more details.

  • macOS: My daily driver for general use and development.

Dev Tools

  • uv: My preferred Python package and project manager. Couldn’t imagine building Python projects without it.

  • ruff: For linting and formatting Python code. Has replaced both flake8 and black in my workflow.