Research Scientist/Engineer, Autonomous Security, DeepMind

About the role

Google DeepMind seeks a Research Scientist/Engineer in autonomous security. Minimums: PhD or equivalent, 4 years training/fine-tuning/evaluating foundation models, 3 years ML or reasoning applied to cybersecurity/systems/program analysis/code generation, and 2 years LLM post-training with SFT, execution-feedback RL, and trajectory reward modeling. Work includes Gemini post-training, autonomous security agents, simulations, benchmarks, and product deployment. Sponsorship is not addressed.

Requirements

• PhD in CS, cybersecurity, ML, related field, or equivalent practical experience
• 4 years Python and ML frameworks training, fine-tuning, and evaluating foundation models
• 3 years ML or automated reasoning applied to cybersecurity, systems, program analysis, or code generation
• 2 years LLM post-training including supervised fine-tuning, execution-feedback RL, and multi-step reward modeling
• Published ML/cybersecurity research or production security AI preferred
• On-site, with preferred location options New York, Mountain View, or San Francisco

What you'll do

• Design post-training methods for Gemini including execution-feedback reinforcement learning and reward modeling
• Build scalable simulated environments and synthetic-data pipelines for complex security workflows
• Architect autonomous agents for multi-step planning and tool use under adversarial conditions
• Create evaluations and benchmarks for offensive, defensive, and vulnerability tasks
• Deploy research into internal defenses and customer products and publish results

Where this was posted

• LinkedIn

  last seen 2h ago
  View on LinkedIn