Senior Security Engineer, Agentic Red Team, DeepMind
Mountain View, CA, USA +2 · Senior · Full-time · Posted Oct 27, 2026
- Location
- Mountain View, CA, USA
New York, NY, USA
Zürich, Switzerland - Experience
- 5 years in Red Teaming, Offensive Security, or Adversarial Machine Learning.
- Education
- Bachelor's degree in Computer Science or Information Security, or equivalent practical experience.
About the job
Conduct adversarial security assessments of agentic AI services, then automate vulnerability testing and help developers implement defenses.
What you'll do
- Assess agentic services for vulnerabilities, including prompt injection, tool-use escalation, and autonomous lateral movement.
- Develop attack sequences and convert discoveries into automated regression tests and reusable attack patterns.
- Work with developers during design and build phases to provide security feedback and defensive guidance.
What they ask for
- Bachelor's degree in Computer Science or Information Security, or equivalent practical experience.
- 5 years of experience in Red Teaming, Offensive Security, or Adversarial Machine Learning.
- Experience with LLM architectures, agentic workflows, and AI vulnerability classes.
- Experience developing exploits for GenAI models, such as prompt injection, adversarial examples, or training data extraction.
Nice to have
- Experience consulting with product teams or working in a startup-like environment.
- Familiarity with AI safety benchmarks, evaluation frameworks, and fuzzing techniques.
Pay
$174000 - $252000 USD base salary
Tools
Python, Go, C++
AI focus
Red-team agentic AI models and services, targeting vulnerabilities in tool use, reasoning, and other agentic behaviors; build automated testing and defensive strategies.
Model-written summary · the original posting is authoritative