Frontier Safety Mitigations Research Engineer, DeepMind

Overview In this role you will build advanced classifiers and data pipelines to detect misuse, and own end-to-end evaluation and rapid iteration. You will create cross-context monitoring to identify large-scale attack vectors and deploy data-driven response systems against persistent actors.

You will evaluate and secure agentic AI systems through threat modeling and robust mitigations, and advance automated red-teaming and adversarial robustness research. This position sits at the frontier of AI safety and product innovation, with impact across research and production in a mission-driven lab.

Responsibilities Build classifiers and data pipelines to detect misuse Develop cross-context monitoring and signal aggregation for detecting large-scale attack vectors Implement data-driven, semi-automated account-level response systems against malicious actors Evaluate and secure agentic AI systems with threat models and testing environments Advance red-teaming and adversarial robustness research using multi-turn/agentic attacks Key requirements Bachelors degree in Computer Science, Machine Learning, or related technical field, or equivalent practical experience 5 years of software development experience including Python Experience with research-to-deployment pipeline in frontier AI Experience working in a software engineering or research team collaboration architectural judgement problem-solving under uncertainty Python adversarial ML threat modeling

Report this job