CBRNE AI Safety Red Teamer

Handshake; United States; Remote; USD 140,000 - 210,000; Full time

Job summary

Handshake seeks a CBRNE Red Teamer to evaluate AI models for safety and security risks across chemical, biological, radiological, nuclear, and explosive domains. You will craft adversarial scenarios, assess model defenses, and document how and why failures occur to guide improvements.

This role blends national security with cutting-edge AI safety, requiring deep domain expertise, ethical judgment, and the ability to work with researchers, policy teams, and engineers to strengthen Handshake AI’s

Qualifications

• Graduate-level education or equivalent in a CBRNE field.
• Ability to evaluate model outputs for technical risk in CBRNE domain.
• Understanding of dual-use research and the distinction between open-source knowledge and operational uplift.
• Hands-on experience with multiple LLMs (ChatGPT, Claude, Gemini, etc.).
• Strong written communication and ethical judgment.
• Self-directed, collaborative, and feedback-ready.

Responsibilities

• Design adversarial prompts testing for uplift toward CBRNE threats.
• Evaluate outputs for technical accuracy and dangerous content.
• Probe dual-use boundaries and open literature vs uplift.
• Test multi-step attack chains simulating threat actor behavior.
• Document findings with rationale and risk significance.
• Collaborate with AI researchers and policy teams to drive improvements.

Skills

CBRNE domain expertise

LLM proficiency

Adversarial problem solving

Technical communication

Ethical judgment

Self-directed / collaborative

Education

Graduate-level education in a CBRNE field

Tools

Python

LLM APIs

Job description

Handshake seeks a CBRNE Red Teamer to evaluate AI models for safety and security risks across chemical, biological, radiological, nuclear, and explosive domains. You will craft adversarial scenarios, assess model defenses, and document how and why failures occur to guide improvements.

This role blends national security with cutting-edge AI safety, requiring deep domain expertise, ethical judgment, and the ability to work with researchers, policy teams, and engineers to strengthen Handshake AI’s