Frontier Safety Mitigations Research Engineer, DeepMind
Frontier Safety Mitigations Research Engineer, DeepMind Overview
| Company Name | |
| Job Role | Frontier Safety Mitigations Research Engineer, DeepMind |
| Qualifications | Bachelor’s |
| Category | IT Jobs |
| Job Type | Full Time |
| Location | London |
DeepMind, a leading AI research organization under Google, is seeking a Frontier Safety Mitigations Research Engineer to join their team in London, UK. This role involves developing advanced safety measures and mitigation strategies for frontier AI systems, ensuring their safe deployment and operation in real-world environments.
Overview
The successful candidate will be responsible for designing and implementing sophisticated classifiers and data pipelines aimed at detecting misuse and malicious activities. They will oversee the entire process from automated model evaluation to rapid iteration, ensuring models remain effective against evolving threats. Additionally, they will develop comprehensive monitoring systems capable of identifying coordinated harmful behaviors across multiple user sessions by aggregating signals to uncover large-scale attack vectors.
Part of the role involves creating semi-automated, data-driven response systems at the account level to identify and counter persistent malicious actors, utilizing rich signals from live production traffic. The engineer will also evaluate and enhance the security of autonomous AI agents by developing threat models, establishing testing environments, and deploying defenses against sophisticated hacking attempts and long-term attack strategies.
Research in automated red-teaming and adversarial robustness will be a key focus, employing multi-turn and agentic attack techniques to systematically uncover vulnerabilities and potential misuse points in AI systems.
Key Responsibilities
- Design and build classifiers and data pipelines to detect misuse, managing the entire lifecycle from automated evaluation to model iteration.
- Develop cross-context monitoring systems to identify coordinated harms, using innovative signal aggregation methods across user sessions.
- Create account-level response mechanisms that leverage production signals to detect and respond to malicious activities.
- Assess and improve the security of AI agents by creating threat models, testing environments, and deploying mitigations against advanced hacking and long-horizon attacks.
- Advance research in automated red-teaming and adversarial robustness, focusing on multi-turn and agentic attack techniques to identify vulnerabilities.
Qualifications
- BachelorĂ¢??s degree in Computer Science, Machine Learning, or a related technical field, or equivalent practical experience.
- Minimum of five years of experience in software development, with proficiency in Python.
- Experience with the research-to-deployment pipeline in frontier AI environments.
- Experience working within a research or software engineering team, demonstrating collaborative skills.
Preferred Skills and Experience
- Background in cybersecurity detection, anomaly detection at scale, and translating safety research into production systems.
- Experience leading or collaborating on applied machine learning projects, especially involving large language models, inference, and fine-tuning.
- Proficiency with AI coding agents, architectural judgment, and hardware accelerators such as TPUs and frameworks like JAX.
- Knowledge of adversarial machine learning, automated red-teaming, model interpretability, and probing techniques.
- Understanding of AI safety control measures, including chain-of-thought monitoring, faithfulness, and system monitorability.
About the Organization
DeepMind is dedicated to advancing AI for the benefit of society, prioritizing safety and ethics in all research endeavors. The organization employs interdisciplinary teams working on a variety of challenging problems, aiming to develop AI solutions that address global issues and accelerate scientific discovery. The role offers opportunities for professional growth, collaboration, and meaningful impact.
Additional Information
Applicants should note that Google emphasizes the importance of English proficiency for global collaboration. The company does not accept unsolicited agency resumes and clarifies that any equity grants are at the sole discretion of Alphabet Inc., with specific terms and conditions.
Google is committed to equal opportunity employment, welcoming applicants regardless of race, gender, sexual orientation, gender identity, disability, age, or other protected characteristics. The company provides accommodations for applicants with special needs upon request.
For more details about privacy policies and legal terms, applicants are encouraged to review Google’s policies on their website.
Degree Requirement: Bachelor’s
Visa Sponsorship Promising
To apply for this job please visit www.google.com.