Build and maintain the infrastructure for large-scale reinforcement learning, enabling researchers to run, debug, compare, and reproduce experiments across distributed training and inference systems.
Vmax jobs
Vmax is developing AI agents for open-ended learning.
7 open roles at Vmax
Develop novel reinforcement learning algorithms for post-training language models, adapting ideas from traditional RL and evaluating sample efficiency, robustness, generalization, and failure modes.
Develop mechanistic interpretability methods that turn language-model representations and behaviors into intrinsic reinforcement-learning rewards, and evaluate how these signals improve training and generalization.
Develop a focused research project at the intersection of mechanistic interpretability and reinforcement learning, using model internals to create and evaluate training signals and contribute to publications.
Develop reinforcement-learning methods and systems that help LLM agents discover objectives, tasks, and curricula, while leading a focused research project and communicating findings through experiments and publications.
Build and improve reinforcement learning training systems for language-model agents, translating research into reliable implementations, evaluations, and infrastructure while diagnosing training issues and measuring model behavior.
Develop reinforcement-learning methods and systems that enable agents to discover their own objectives, tasks, and curricula, while leading empirical research on open-ended learning.
Page 1 Β· 7 results
