[Remote] Research Engineer, Security and Privacy, DeepMind

Note: The job is a remote job and is open to candidates in USA. Google DeepMind is a pioneering AI lab focused on advancing AI development to solve complex global challenges. The Research Engineer role involves building robust evaluation systems to measure model vulnerabilities and collaborating with teams to enhance model performance.

Responsibilities

• Build robust pipelines, benchmarks, and auto-red teaming tools to measure model vulnerabilities and adversarial robustness
• Collaborate closely with post-training teams to turn your evaluation insights and discovered failure modes into actionable model upgrades
• Generalize your solutions into reusable evaluation libraries and frameworks for protecting agents across Google, and sharing your knowledge through publications, open source, and education

Skills

• Bachelor's degree or equivalent practical experience
• 5 years of experience with software development in one or more programming languages
• 3 years of experience testing, maintaining, or launching software products, and 1 year of experience with software design and architecture
• 1 year of experience in evaluating, stress-testing, or red-teaming generative models to understand their capabilities, limitations, and security flaws
• Master's degree or PhD in Computer Science or related technical field
• Experience or publications in ML security, adversarial robustness, privacy, safety, or alignment
• Research experience or publications in ML security, adversarial robustness, privacy, safety, or alignment
• Ability to leverage Python and frameworks like JAX or PyTorch to build robust evaluation systems, adapting internal tooling shifts
• Ability to autonomously translate abstract adversarial research into concrete benchmarks
• Ability to drive dissect frontier models

Benefits

• 15% bonus target
• Bonus
• Equity
• Benefits

Company Overview