Blog

Notes on machine learning, neuro-symbolic reasoning, data-protection engineering, and the systems that have to hold all of it up. Mostly things I wish someone had written down before I started.

A three-by-three matrix of abstract glyphs with the final cell left empty

Reproducing a neuro-symbolic model for Raven's Progressive Matrices

Six months of trying to rebuild a published three-stage reasoning pipeline from the paper alone. What the paper said, what it left out, and where the accuracy actually came from.

2 min read

LLM inference that never leaves the building

Classifying survey questions against Article 9 GDPR means the data is, by definition, the data you are not allowed to send anywhere. Notes on building the whole thing on-premise.

2 min read

Three stacked containers inside a closed network boundary, with the route to the internet blocked

This blog is just files

How the site you are reading works: mdsvex posts in a folder, a glob, and a full prerender onto Cloudflare's edge. No database, no CMS, no admin login.

2 min read

A stack of markdown source files fanned out beside a rendered page