About

CoDHe Labs is a personal, open-source space where I share work at the intersection of AI and chemistry — built in my own time and at my own expense because I believe openly published code and methods matter.

Who I am

I’m Daniel Halwell, MChem. I’m an AI Engineer (Platform Developer) at Synextra, based in the north west of the UK, where I help build agentic AI applications and platform capabilities. My professional work is with Synextra; this site and CoDHe Labs are not sponsored by my employer and do not represent their views unless explicitly stated.

Background

I studied chemistry at Loughborough University. Before moving into a full-time AI role, I worked as an analytical chemist at Sanofi, Recipharm, and AstraZeneca — work that shaped how I think about measurement, method development, and what “good evidence” looks like in the lab.

That path informs how I approach ML for chemistry: I care about reproducibility, honest limits of models, and tooling others can actually run.

What CoDHe Labs is for

CoDHe Labs is where I intend to publish open-source research — notes, experiments, and software — spanning topics I’m exploring in AI and chemistry. It may grow into preprints, papers, or community tools; the through-line is open publication so others can inspect, reuse, and improve the work.

How this project shows up (brand character)

I use Jungian brand archetypes the way strategists do: pick a dominant personality so the voice stays consistent, and a smaller accent so it doesn’t feel one-note. For CoDHe Labs that mix is roughly 70% Sage / 30% Creator — same idea as the “core + influencer archetype” split in references like Mark & Pearson’s The Hero and the Outlaw.

Sage ( desire: Understanding) — Prefer clear explanations, sources you can follow, and honest limits. The point is insight others can reuse — not performative novelty.

Creator ( desire: Innovation) — Treat code and experiments as first-class artefacts: rough edges are acceptable if they make the idea inspectable.

Practically: long-form and research pages should read precise and well sourced; repositories and demos can be a bit more expressive and experimental — as long as the thread back to evidence stays obvious.

Open source

I publish open source because transparency and shared tooling accelerate science. If you use something from this site, a citation or link back is welcome when it makes sense in your context.

Contact

For CoDHe Labs specifically: research@codhelabs.co.uk

Add “CoDHe Labs” or the project name in the subject line so messages are easy to triage.