Research Scientist, Gemini Safety and Behavior

Highlights

• Drive jailbreak mitigations deployable at scale across Gemini products.
• Build red and blue teaming for text, multimodal and agentic models.
• Improve adversarial robustness for high-stakes abuse in agentic settings.

What they offer

• 20% bonus target
• equity
• benefits

Ideal candidate

Applied researcher with strong experimental judgment who cares about AI safety and alignment, eager to take ideas from hypothesis to shipped mitigations in a fast, collaborative team.

Requirements

• PhD in Computer Science, related field, or equivalent practical experience
• 2 years of experience in Large Language Model safety or security
• Experience developing and leveraging agentic workflows around safety, behavior, and alignment
• Experience with synthetic data generation pipelines and evaluations for non-verifiable tasks
• Experience taking research from concept to product
• Experience collaborating or leading an applied research project
• Strong experimental judgment on baselines, ablations, and testing priorities
• Track record of publications at NeurIPS, ICLR, ICML