Essays and thoughts on AI, machine learning, and building things that matter.
A setup nightmare, a lavender plant that's still alive, and a Llama 4 confession. An honest take on whether Meta's AI glasses are worth it — and what it would take for them to actually be great.
Read more →Most AI projects don't fail at the model — they fail at the integration. Here's how I work as an embedded FDE: production code from day one, owning deployment, bridging product, customer, and execution.
Read more →After 10+ years in AI/ML — a PhD, publications at ICML, ICCV, and ICDAR, and building production AI systems — I've founded Probably Approximately Human, a consultancy BV based out of Belgium.
Read more →