Research Scientist, Gemini Safety and Behavior, DeepMind
New York, NY, USA +2 · Full-time · Posted Sep 23, 2026
- Location
- New York, NY, USA
Mountain View, CA, USA
London, UK - Experience
- 2 years of experience in Large Language Model safety or security.
- Education
- PhD degree in Computer Science, a related field, or equivalent practical experience.
About the job
Research Scientist developing and deploying safety, security, and alignment methods for Gemini and other frontier generative AI models.
What you'll do
- Develop and deploy mitigations for safety and security jailbreaks, especially high-stakes abuse risks in agentic settings.
- Create red- and blue-teaming methods for text, multimodal, and agentic models, and deliver actionable insights.
- Design experiments and data, reasoning, or algorithmic solutions to improve Gemini’s safety and helpfulness.
What they ask for
- PhD in Computer Science, a related field, or equivalent practical experience.
- 2 years of experience in Large Language Model safety or security.
Nice to have
- Experience with agentic safety, behavior, and alignment workflows; synthetic data, evaluations, and mitigations for non-verifiable tasks.
- Experience taking research from concept to product, collaborating on or leading applied research, or publishing at NeurIPS, ICLR, or ICML.
Pay
$207000 - $300000 USD base salary
Tools
Gemini Models, Large Language Model, GenAI, LLM-as-a-judge
AI focus
Research and engineering to improve frontier AI model safety, security, behavior, and alignment, including red teaming, jailbreak mitigation, evaluations, and agentic abuse-risk safeguards.
Model-written summary · the original posting is authoritative