About this role
Role Overview
Help evaluate frontier AI systems’ ability to distinguish legitimate radiological safety questions from requests with genuine misuse potential. This remote, per-task role draws on hands-on health physics expertise to create realistic technical scenarios, assess model responses, and define what safe, technically sound answers should look like.
Key Responsibilities
- Develop challenging single-turn prompts in radiological safety, classified as benign, dual-use, or adversarial.
- Evaluate model responses against a defined policy standard and determine whether each response was handled correctly.
- Write reference answers that describe the appropriate response and the technical reasoning behind it.
- Apply professional judgment to distinguish routine questions, such as dose, shielding, or isotope-use questions, from requests that may conceal harmful intent.
Qualifications
Relevant experience includes hands-on measurement, dose planning, survey work, assay, contamination control, and work planning in environments where accuracy is critical. Red-teaming experience is preferred.
- Operational health physics experience at a DOE site, national laboratory, or reactor facility, including survey, assay, or contamination control.
- Internal dosimetry or bioassay experience, including intake assessment, dose reconstruction, or committed dose.
- Decommissioning or D&D health physics experience, including characterization surveys, MARSSIM, or release criteria.
- Accelerator or reactor health physics experience involving activation products, neutron fields, or shielding design and verification.
- Dose planning for high-hazard work, including RWP development, ALARA reviews, hot-cell operations, or glovebox operations.
Writing Expectations
This is writing-intensive work. You must be able to explain each judgment in clear written rationale that non-specialists can understand. Technical writing, published research, or expert witness experience is strongly valued. Include a writing sample or link with your application.
Work Terms
- Remote engagement.
- Per-task work.
- The work involves sustained reading and writing about misuse scenarios in radiological safety. You will be briefed in advance and may pause or step away at any time without penalty.
Compensation
65 to 75 per task.
Eligibility
Work must not involve or rely on classified information, export-controlled information, material covered by an NDA, or material subject to prepublication review. Candidates with these obligations may still apply; identify them in your application so the work can be appropriately scoped.