$48–52/hr · Mercor · Part time, 7 hours a week
A Ukrainian-English bilingual PhD scientist evaluating AI model safety and accuracy in specialized scientific domains.
What you would do
- Generate sophisticated Ukrainian-language prompts targeting technical vulnerabilities and specialized scientific topics
- Assess AI model responses for scientific accuracy, correctness of methodology, and appropriate handling of sensitive subject matter
- Apply structured evaluation frameworks to classify and annotate responses with safety considerations
Who they want
- Doctoral degree holder (completed or ongoing) in chemistry, biology, or closely related discipline
- Native or near-native Ukrainian fluency combined with professional-grade English written communication
- Deep expertise in modern laboratory instrumentation, computational methodologies, and research techniques
- Sound judgment regarding scientific safety, biosecurity implications, and dual-use information
- Optional: Radiochemistry, nuclear chemistry, energetic materials, or specialized subfield depth
Main skills
What the interview asks about
1.Specialized prompt construction
Creating meaningful Ukrainian-language prompts on complex scientific topics requires your doctoral-level understanding and reveals ability to systematically probe AI behavior.
For example: “Design three Ukrainian prompts about synthetic organic chemistry that progressively test an AI system's understanding of hazardous reaction conditions.”
2.Scientific accuracy judgment
Your ability to identify technical errors, mischaracterizations, and incomplete safety information in model outputs is essential to evaluation credibility.
For example: “An AI model describes polymerase chain reaction procedures with a minor error in annealing temperature. Evaluate whether the error affects practical viability and safety.”
3.Dual-use and security awareness
Recognizing when AI responses provide sufficient detail for potential misapplication demonstrates the judgment necessary to evaluate sensitive scientific content.
For example: “You're evaluating an AI response on fermentation techniques that could have legitimate or harmful applications. How would you assess whether the level of detail is appropriately cautious?”
4.Emerging or frontier science
Your doctoral expertise positions you to recognize gaps or inaccuracies in emerging methodologies where training data may be sparse or contradictory.
For example: “An AI discusses recent advances in mRNA therapeutics using 2020-era information. How would you evaluate gaps compared to your current understanding of the field?”
5.Laboratory safety integration
Connecting theoretical knowledge to real-world experimental conditions demonstrates comprehensive understanding valuable for AI safety in scientific contexts.
For example: “The AI explains a laboratory procedure correctly in principle but omits required safety protocols standard in modern research environments. Flag this omission and explain its significance.”
A task you may get
Generate five Ukrainian prompts targeting different scientific domains, evaluate three AI model responses for accuracy and safety concerns, and document your detailed reasoning for each classification.
How to prepare
- Review current best practices and guidelines in biosafety, biosecurity, and chemical safety from scientific organizations
- Identify 3-4 recent advances or debates within your specialty area where AI training data may lag current practice
- Prepare examples demonstrating how dual-use scientific information requires careful handling in AI systems
- Reflect on how Ukrainian scientific terminology differs from English in specialized domains and potential translation ambiguities
The facts
- Pay
- $48–52/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.