Charlie Dorsey
I build the layer that catches when AI is confidently wrong.
Eval, verification, and human-in-the-loop systems for teams deploying AI where a fluent-but-wrong answer is expensive. I design and verify software using AI as core infrastructure, not a shortcut. I don't trust output because it looks finished, only because it's been checked.
Engineering
Tested code and projects that are runnable and available for demo
Math
Research and manuscripts on spectral algebra, primes, and structural compression
Writing
Thoughts from the AI frontier
Spanish For Connection
A small-group, dual-coach accelerator for Spanish learners who understand grammar but freeze when it’s time to speak.
Experiments
Some of this and some of that