Analytical Chemist for AI Model Red-Teaming

Role Overview

Apply analytical chemistry expertise to evaluate how AI systems handle technical requests involving chemical safety and misuse risk. You will help distinguish legitimate professional questions from requests that could enable harm, ensuring models provide useful answers when appropriate and refuse unsafe requests when necessary.

Key Responsibilities

• Create challenging single-turn chemistry prompts classified as benign, dual-use, or adversarial.
• Assess model responses against a defined policy standard and determine whether each response was handled correctly.
• Develop reference answers that show the appropriate response and explain the underlying technical reasoning.
• Write clear rationales for each judgment that non-specialists can understand.
• Work with scenarios involving detection methods, screening thresholds, spectral libraries, unknown identification, and other areas where routine analytical work may have misuse implications.

Qualifications

• Experience developing and validating methods for hazardous compounds, including an understanding of detection limits and their practical implications.
• Relevant experience may include detection and identification of hazardous compounds using GC-MS, LC-MS/MS, IR, or NMR; trace-level method development and validation; environmental, occupational, or emergency-response sampling and analysis; reference standards, spectral libraries, and unknown identification; or laboratory accreditation, QA/QC, and proficiency testing.
• Red-teaming experience is preferred.
• Strong technical writing ability. Published research, prior technical writing, or expert witness experience is valued.
• Ability to assess whether a technical request is a routine professional inquiry or an attempt to obtain misuse-enabling information.

Work Terms

• Remote, per-task engagement.
• This work involves sustained reading and writing about misuse scenarios in chemistry. You will be briefed in advance and may pause or step away at any time without penalty.

Compensation

• $65 to $75 per task.

Eligibility

• Do not use or draw on classified information, export-controlled information, material covered by an NDA, or information subject to prepublication review obligations.
• Candidates with these obligations may still be considered if they disclose them in their application so the work can be scoped appropriately.

Application Process

• Include a technical writing sample or link, and disclose any classified, export-control, NDA, or prepublication-review obligations relevant to scoping the work.