Research Scientist, Safety Oversight

In this role, you'll develop safety monitoring systems to detect issues in deployed AI models using production data and automated evaluation.
Build classifiers and data pipelines to identify model misbehavior and misuse at scale.
Research cross-context monitoring systems to detect coordinated harms and large-scale attack vectors.
Design novel detection methods using model activations, actions, and decision traces.
Collaborate with infrastructure and data science teams to scale safety monitoring and share results.

Desbloqueie com o Pro