Veylan Solmira

AI Safety Researcher

I work on detecting deceptive behavior in frontier models, scalable oversight, and the moral status of digital minds. I was an Astra Fellow (Jan–Apr) working with Apollo Research on scheming-detection evaluations. I'm now extending that work through a FIG Extension Fellowship on persona dynamics.

Featured

FIG Extension Fellowship

Studying persona dynamics in frontier models.

Projects

AI Safety Research Compiler An interactive curriculum for AI safety research.
Art Exploration Music recommendation engine and other experiments.
Earth Explorersoon Interactive travel + trip-planning map.
Drift Taxonomy Surveysoon Cross-locus survey of AI drift mechanisms.

Elsewhere

GitHub· Substack· LinkedIn· Letterboxd· Email