Conduct research at the intersection of mechanistic interpretability and reinforcement learning. Develop methods to analyze language-model and agent internals, derive intrinsic rewards and training signals, run experiments, evaluate failure modes, and build reproducible research infrastructure. The fellow will own a focused project, collaborate with technical staff, communicate findings, and contribute to publications.

Conduct research on open-ended learning for LLM-based agents. Develop reinforcement learning methods for discovering objectives, tasks, curricula, and reward signals; design automated environments, self-play, and intrinsic motivation systems; and study methods to prevent degenerate objectives. Fellows own focused research projects, run experiments, communicate findings, and contribute to publications while collaborating with Vmax technical staff.

Vmax seeks exceptional talent to contribute to an applied research lab developing AI systems capable of open-ended learning. The role is intended for candidates whose strongest contribution may not fit existing positions but who demonstrate unusually strong ability, agency, and ambition. Applicants should show clear evidence of exceptional potential to help advance AI systems beyond the limits of learning from human expertise.