Senior Red Team Security Engineer, AI Agentic

Google LLC; Mountain View, New York (CA, NY); On-site; USD 174,000 - 252,000; Full time

Benefits offered by this job

Equity

Benefits

Job summary

Google DeepMind is seeking a Senior Security Engineer for the Agentic Red Team. You will be the primary technical executor of adversarial engagements on production-scale AI models, focusing on agentic behaviors like tool usage and reasoning chains.

You will help codify findings into reusable guardrails for Google's agent developers. Join a mission-driven team that builds security tooling and frameworks to close the launches gap between capability and safety, contributing to automated red-teaming

Qualifications

• Bachelor's degree or equivalent practical experience in CS/InfoSec.
• 5 years of Red Teambuilt/Offensive Security/Adversarial ML experience.
• Experience with GenAI vulnerabilities and LLM architectures.

Responsibilities

• Conduct rapid security assessments on agentic services focusing on Generative AI vulnerabilities such as prompt injection and tool-use escalation.
• Engineer and execute attack sequences exploiting non-deterministic model behaviors and data poisoning vectors.
• Develop automated regression testing frameworks (Auto Red Teaming) to prevent regressions in future model versions.
• Collaborate with developers during design and build phases to provide immediate feedback and tighten defense.
• Maintain a library of agent-specific attack patterns to establish release criteria for new models.

Skills

Red Teaming

Offensive Security

Adversarial ML

LLM architectures

Prompt injection

Education

Bachelor's degree in Computer Science, Information Security, or equivalent

Tools

Python

Go

C++

Job description

Google DeepMind is seeking a Senior Security Engineer for the Agentic Red Team. You will be the primary technical executor of adversarial engagements on production-scale AI models, focusing on agentic behaviors like tool usage and reasoning chains.

You will help codify findings into reusable guardrails for Google's agent developers. Join a mission-driven team that builds security tooling and frameworks to close the launches gap between capability and safety, contributing to automated red-teaming