Independent work

AI research and development

AI is a sustained personal interest as well as part of my product work. I learn by building: applying it to scientific literature, testing mathematical ideas, and examining how we can check the behaviour of AI-generated software.

← Back to my product and leadership work

Featured project

UpBeam Science

A research reflection network spanning approximately 10,000 scientific articles.

UpBeam Science is my main independent AI project. It explores how ideas across scientific literature connect, bringing individual articles into a wider network for research and reflection.

The project brings together my interests in software, scientific inquiry and the practical uses of AI. It gives me a sustained setting in which to explore what these tools make possible and where careful judgment is still needed.

Get in touch about UpBeam Science →

Further experiments

Questions I explore by building

AI Science

AI-assisted mathematical research

An experiment in faster exact network checks, with an interactive comparison against an exact reference method and inspectable arguments and code.

These are preliminary research results; novelty and independent review remain open.

Explore AI Science →
Hegelese

Checking changes proposed by AI

An experimental functional language and finite-process checker exploring semantic drift in AI-assisted programming. It checks whether a proposed rewrite preserves explicitly stated behaviours and can return a counterexample when a preservation claim fails.

This is a narrow, executable check, with clear limits on what it can establish.

Code and runnable examples ↗

Associated research →