Research Scientist, AI Safety and Security

Position Overview

Google is seeking a Research Scientist to work on AI safety and security, joining a dedicated research team focused on proactive threat mitigation, adversarial machine learning, and safeguarding intelligent agents. The role concentrates on foundational machine learning approaches in AI safety, including continual learning and interpretability, to build robust and trustworthy AI systems.

Primary Responsibilities

• Conduct pioneering research on machine learning techniques such as model robustness, continual learning, interpretability, and multi-objective optimization to advance trustworthy AI technologies.
• Develop and implement thorough evaluation frameworks, scenario-based benchmarks, and stress-testing methods to measure frontier AI capabilities and consensus among multiple agents.
• Create and manage sophisticated datasets and carry out fine-tuning or optimization studies to bolster model resilience against newly emerging threats and uphold safety standards.
• Collaborate extensively with engineering centers, core product teams, and academic collaborators to transform theoretical research into dependable production-grade solutions.
• Publish leading-edge research findings in prominent machine learning conferences and actively engage with academic and industry research communities.

Required Qualifications

• PhD in Computer Science or a relevant discipline, or equivalent practical experience.
• Expertise in machine learning areas, including but not limited to supervised and unsupervised learning, reinforcement learning, adversarial machine learning, safety and robustness, interpretability, generative models, agentic AI, and multi-objective optimization.
• Documented scientific publications submitted to top conferences or journals such as CVPR, ICCV, NeurIPS, ICML, or ICLR.

Preferred Qualifications

• Proficiency in general-purpose programming languages, particularly Python.
• Experience investigating emerging technical AI threats like automated scams, deepfake production, or vulnerabilities in autonomous agents, along with designing effective defense mechanisms.
• Strong expertise in adversarial machine learning fields including AI agent security, data poisoning, prompt injection attacks, and model backdoor detection.
• A rigorous security mindset applied to AI including debugging complex machine learning failures, reverse-engineering model behaviors, and red-teaming cutting-edge AI systems.
• First-authored papers in top-tier machine learning, safety/security, or human-computer interaction conferences.

Additional Information

Applicants who currently hold the legal right to work in Singapore and do not require visa sponsorship from Google will be given preference.

Google is an equal opportunity employer committed to fostering a diverse and inclusive workplace. Hiring decisions are made irrespective of race, color, ancestry, religion, gender, nationality, sexual orientation, age, marital status, disability, gender identity, veteran status, or criminal history, consistent with legal requirements.

If accommodations are necessary for applicants with disabilities or special needs, Google provides support through a formal accommodation request process.