About

I’m Ryan Mulligan. I build research‑grade systems that make model behavior inspectable and testable. My work sits at the intersection of mechanistic interpretability, activation steering, and long‑horizon agent infrastructure.

I’m currently focused on the introspection gap, when a model acts on an internal state it won’t admit, and what that means for evaluation integrity and alignment.

If you work on interpretability, evaluation, or alignment, I’d love feedback or collaboration. If you are interested in sponsoring this research or hiring me, I am open to conversations.