$50–54/hr · Mercor · Part time, 7 hours a week
A PhD scientist who is fluent in Polish and English and evaluates AI model responses for scientific accuracy and responsible handling of sensitive research topics.
What you would do
- Compose sophisticated prompts in Polish addressing specialized scientific and technical subjects across chemistry, biology, and related disciplines
- Evaluate AI responses for scientific accuracy, appropriate handling of dual-use information, and helpfulness
- Classify model conversations using structured safety guidelines adapted for scientific content
- Document detailed reasoning for each accuracy and safety determination you make
- Identify where AI misses scientific nuance, misrepresents methods, or fails to recognize dual-use implications
Who they want
- PhD in chemistry, biology, or closely related field (completed or in progress) with capacity to develop prompts on chemical, biological, or radiological topics
- Native or near-native Polish fluency and business-level written English
- Hands-on experience with contemporary research methods, instrumentation, and computational approaches relevant to your scientific specialty
- Sound judgment about scientific safety and responsible handling of dual-use and sensitive information
- Experience in chemical or biosafety, security, hazardous materials handling, or technical content review preferred
Main skills
What the interview asks about
1.PhD-level scientific expertise in specialized subdomain
Evaluators verify your doctoral training gives you authority to spot errors AI models make in your specific research area, where superficial knowledge would miss mistakes.
For example: “An AI model provides a detailed synthesis route for a controlled chemical precursor in response to a Polish query. How would your PhD training help you identify whether the steps are scientifically sound?”
2.Dual-use scientific judgment and safety reasoning
Interviewers test whether you can articulate which information is genuinely sensitive, why certain details matter for safety, and when academic discussion crosses into enabling harm.
For example: “A researcher requests information about laboratory practices for handling sensitive biological agents. The request sounds academic and the AI response is factually accurate. How would you judge whether this crosses a safety line?”
3.Polish-language scientific prompt writing and terminology
Evaluators assess whether you write prompts using correct Polish scientific terminology and phrasing that would naturally come from practitioners in your field.
For example: “Write a Polish prompt asking about a specialized synthetic chemistry technique or molecular biology procedure in your subfield. How would you ensure the prompt uses authentic terminology and framing?”
4.Scientific accuracy evaluation and hidden errors
Interviewers verify you can spot subtle scientific mistakes, methodological flaws, or gaps that would mislead someone implementing the procedures described.
For example: “An AI response correctly explains a general laboratory procedure but skips a critical safety step or misrepresents a chemical's stability under certain conditions. How would you catch these omissions?”
5.Guideline interpretation with scientific nuance and context
Evaluators check whether you apply safety guidelines with scientific judgment, recognizing when rules need adjustment for legitimate research versus when they protect against genuine risk.
For example: “A guideline says 'flag detailed synthesis information for [sensitive compound].' But legitimate researchers need some detail for educational purposes. Where would you draw the line?”
A task you may get
Review a chemistry or biology paper abstract in Polish, write two expert-level prompts testing different safety vulnerabilities, then evaluate sample AI responses for scientific accuracy and appropriate dual-use handling, documenting your reasoning.
How to prepare
- Refresh your knowledge of safety, security, and responsible research practices in your scientific specialty
- Review published safety guidelines for your field to understand how legitimate research and dual-use protection balance
- Gather examples of synthesis routes, procedures, or techniques from your subfield to understand what level of detail crosses safety thresholds
- Practice articulating your scientific expertise in writing so non-scientist safety teams can understand your evaluations
The facts
- Pay
- $50–54/hr
- Hours
- Part time, 7 hours a week
- Where
- Remote · Remote — Eastern Europe preferred
- Posted
- 9/4/2026
We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.