LUMENIA studio

This site is a set of working notes on AI engineering: how systems are built, how they fail, and how to measure them. The scope is not fixed. Seven building blocks form the starting sequence: vector representation, function approximation, attention, probability, context, evaluation, and the agent loop. More topics follow as the work demands.

The building-block pages state a thesis, cite a paper, a book, and a course, show where the primitive is used, and define a lab that can be reproduced from a repository. The patterns and labs exist so those claims can be inspected.