Alignment Research jobs

Alignment research is the work of making capable models want what we meant: reward modeling, RLHF and constitutional-style methods, scalable oversight, and the theory of what training actually optimizes. Most seats sit at frontier labs, with a smaller ring at nonprofits and university-adjacent groups.

The field prizes people who can turn a philosophical worry into an experiment with a metric by Friday. Empirical work dominates hiring — training reward models, building oversight protocols, studying deceptive behavior in controlled settings. Openings are few and applications many, but published independent work still genuinely moves candidates.