Remote Radiological Security Specialist for AI Safety

Mercor; San Francisco (CA); Híbrido; GBP 67.000 - 78.000; Jornada completa

Descripción de la vacante

Mercor connects elite creative and technical talent with leading AI research labs. This remote contract role is Source Security Specialist offering $65-$75/hour.

You will write challenging prompts in radiological safety, evaluate AI responses, and collaborate with experts to distinguish routine from dangerous queries, all while working independently to meet deadlines and improve model performance.

Formación

• Experience with security or vulnerability assessments of licensed facilities holding Category 1 or 2 sources.
• Expertise in physical protection design and review, including insider threat considerations.
• Background in source recovery, consolidation, or security-upgrade programs.
• Familiarity with regulatory security inspection and compliance (e.g., 10 CFR 37).
• Knowledge of transport security for radioactive material.

Responsabilidades

• Write challenging single-turn prompts in radiological safety, labeled as benign, dual-use, and adversarial.
• Evaluate AI model responses against a policy standard to ensure correct handling.
• Write reference answers with technical reasoning for correct responses.
• Collaborate with experts to draw reliable lines between routine and dangerous queries.
• Work independently and asynchronously to meet deadlines and improve AI model performance.

Conocimientos

Security assessment

Physical protection design

Insider threat

Source recovery/consolidation

Regulatory compliance (10 CFR 37)

Transport security for radioactive材

Descripción del empleo

Mercor connects elite creative and technical talent with leading AI research labs. This remote contract role is Source Security Specialist offering $65-$75/hour.

You will write challenging prompts in radiological safety, evaluate AI responses, and collaborate with experts to distinguish routine from dangerous queries, all while working independently to meet deadlines and improve model performance.