Welcome to limi-lab

Limi-lab exists because some kinds of work do not arrive neatly packaged as papers, projects, or finished arguments. Sometimes the thing comes first as a pattern: a repeated behaviour, a pressure in a system, a question that keeps returning, or a practical problem that refuses to stay merely practical.

I’m Immie. This is my research space. This site is where I keep those things visible.

The work here sits around adaptive AI systems, multi-agent environments, synthetic evaluation tasks, and the question of what stability means when a system is not static. I am especially interested in configurations that can change without collapsing, return without becoming rigid, and differentiate without being forced into sameness.

This is not an academic archive. It is not a product page. It is not a claim that everything here is settled.

It is a lab book.

That means some pages are clearer than others. Some are current summaries. Some are provisional notes. Some are kept because they mark a useful uncertainty. A few may later turn out to be wrong in productive ways.

The aim is not to make the work look finished before it is ready.

The aim is to make it returnable.