code.

My more comprehensive GitHub profile can be found here. A lot of my recent code is tied to private ventures, but the public repositories give a decent sense of the kinds of things I build.

In healthcare I build practice management systems that span the entire arc of a specialist practice, from front-office scheduling and patient intake, through clinical documentation and AI scribes, to billing and revenue cycle management, all augmented by intelligent systems, whether AI or plain automation. These are production platforms used every day by tens of thousands of specialists and the patients they care for.

In biotech that means cell-assay platforms that run ML over single-cell data and model the cost of a protocol, structured databases for managing lab protocols and experimental data, and agentic systems that digest, retrieve and synthesize the biomedical literature. I built and maintain a cardiac drug discovery platform spanning functional assay, proteomic, genomic and biobank data, which is powering multiple cardiac drug discovery labs in Australia.

On the hardware side I’ve written controllers and graphical interfaces that drive liquid handlers and microfluidic cell-culture circuits over many days of continuous, slow flow.

In geospatial I’ve built toolkits for querying dozens of open Australian datasets at once (imagery, demographics, property history, transport, flooding) and layered predictive models on top.

And in astrophysics, my earlier work produced one of the largest, highest-resolution suites of Milky Way-sized cosmological simulations, along with the no-code tooling to design and run those “zoom-in” simulations on HPC clusters. This grew out of studying cosmology and galaxy formation at the Kavli Institute at MIT, where I led the Caterpillar Project, and contributing to the Illustris Project at the Center for Astrophysics at Harvard University.

I also experiment with a homelab on various hardware (Unraid, TrueNAS, bare-metal Kubernetes and more), where I self-host services, deploy quantised LLMs and pipelines locally, and tinker with a Raspberry Pi-driven Home Assistant setup (with mixed results).