AI Red Teamer, CBRNE

This is a fully remote position, open to applicants in United States.

📋 Description

• Assess the capability of AI models to effectively handle inquiries concerning chemical, biological, radiological, nuclear, and explosive threats.

• Develop technically sound adversarial prompts to test if models enhance responses related to CBRNE threats.

• Analyze model outputs for technical precision and the presence of hazardous information.

• Investigate dual-use knowledge limits in relation to scientific, medical, industrial, and weapon applications.

• Execute tests on multi-step and multi-turn attack sequences.

• Evaluate model responses using structured harm taxonomies and severity criteria.

• Record findings with clear and logical technical justifications.

• Differentiate between publicly available literature and genuine operational enhancements.

• Contribute to the development of CBRNE-specific evaluation frameworks and threat models.

• Collaborate with red team members, AI researchers, and policy teams to translate insights into model enhancements.

• Keep abreast of developments in model capabilities, jailbreak methods, and relevant domain advancements.

⛳️ Requirements

• A graduate-level education or equivalent professional experience in a relevant CBRNE discipline (such as chemistry, biochemistry, microbiology, virology, nuclear physics, radiochemistry, materials science, munitions/ordnance, chemical engineering, or closely related fields).

• Capability to assess the technical accuracy and real-world implications of model outputs within your area of expertise.

• Knowledge of dual-use research issues and the differentiation between open-source knowledge and operationally significant enhancements.

• Extensive hands-on experience with various LLMs (including ChatGPT, Claude, Gemini, open-source models, etc.).

• Innovative and adversarial problem-solving abilities.

• Excellent written communication skills, with the ability to convey technical risks to non-specialist audiences clearly.

• Strong ethical discernment and the capacity to separate adversarial thought from personal beliefs.

• Self-motivated, collaborative, and comfortable in environments that require feedback.

• Candidates should be able to engage with sensitive CBRNE-related content in a professional and sustainable manner.

• An active or previous security clearance (Secret, Top Secret, or SCI) is a plus.

• Experience in threat assessment, WMD analysis, intelligence analysis, or arms control verification is advantageous.

• A background in biosafety/biosecurity, chemical safety, nuclear nonproliferation, or explosive ordnance disposal is beneficial.

• Familiarity with relevant regulatory frameworks, such as CWC, BWC, IAEA safeguards, ATF regulations, and Export Administration Regulations, is a plus.

• Experience in red teaming, penetration testing, or structured adversarial evaluations is advantageous.

• Knowledge of Python or scripting languages, LLM APIs, or evaluation tools is a plus.

• Published research or professional presentations in a relevant CBRNE field is advantageous.

• Previous experience in trust and safety, content moderation, or AI evaluation is a plus.

🏝️ Benefits

• Comprehensive health and wellness programs.

• Opportunities for professional development and training.

• Flexible working arrangements to promote work-life balance.

• Engaging team culture with collaborative projects.

• Competitive salary and performance-based incentives.