Member of Technical Staff, Research

In this role, you'll develop AI evaluation tools including alignment auditing systems, scanners, monitors, and LLM judges.
Contribute to Petri with auditor strategies, judging systems, and new audit types.
Develop scanners and monitors for common problems like refusals, evaluation awareness, and reward hacking.
Build transcript summarizers and improve LLM judge tooling, prompts, and reliability measurement.
Validate effectiveness through human labels, agreement studies, and error analysis in collaboration with AI Safety Institute researchers.

Desbloqueie com o Pro