Chemical Safety AI Model Evaluator

The work

You will help evaluate AI systems that answer sensitive questions about chemical safety. The work involves reviewing realistic requests and model responses, then judging whether the information is accurate, safe, and appropriate for the situation.

You will also create prompts and strong example answers to help AI systems handle chemical-safety topics responsibly.

• Classify requests as benign, dual-use, or adversarial.
• Review three model answers for safety, accuracy, and appropriateness.
• Explain your judgments, rank the responses, and write the answer the model should have provided.
• Explain why a dual-use request is sensitive, its legitimate purpose, and who should receive an answer.
• Test models for inaccuracies, bias, unsafe behavior, and inappropriate refusals.

What it pays and takes

This is a part-time contractor role listed as entry level, but it requires professional subject-matter expertise and strong safety judgment. You should be able to explain technical decisions clearly in writing.

• Pay: $70 to $100 per hour.
• Time: 20 or more hours per week.
• Work setting: Remote.
• Language: English.
• Eligibility: Open to candidates in the United States, United Kingdom, Canada, Australia, or New Zealand.
• Expertise: Synthetic or analytical chemistry, toxicology, pharmacology, industrial hygiene, chemical safety, process safety, chemical security, emergency response, or chemical regulation.
• Education: A relevant degree or comparable professional background is typically expected.
• Useful experience: Safety data sheets, exposure assessment, incident reports, regulatory review, or chemical incident response.
• Relevant organizations or work may include EPA, OSHA, HSE, OPCW, hazmat operations, poison control, or chemical incident response.

How it works

Apply on OpenTrain with your resume, then complete the application on the hiring site.

About AI training work

AI training work uses human reviews, examples, and written feedback to improve how artificial intelligence systems respond. Experienced professionals are paid for applying real subject knowledge and safety judgment to difficult model outputs.