$48–52/hr · Mercor · Part time, 7 hours a week
A doctoral-level STEM scientist bilingual in Russian and English who safeguards AI models by evaluating their responses to specialized technical queries.
What you would do
- Generate expert-caliber Russian queries addressing chemical synthesis, biological processes, radiological topics, or nuclear chemistry to probe model comprehension
- Scrutinize generated replies for scientific accuracy, alignment with contemporary laboratory standards, and computational methodology validity
- Assess whether outputs disclose hazardous procedural details, biosecurity gaps, or dual-use applications irresponsibly without adequate safeguards
- Tag conversation patterns against safety guidelines using structured classification systems
- Document assessment findings with disciplinary justification and identified safety implications
Who they want
- PhD-holder in chemistry, molecular biology, physics, radiochemistry, or associated science disciplines, either completed or ongoing
- Native-level or near-native Russian fluency paired with competent business English for written and spoken work
- Extensive laboratory research or computational modeling experience demonstrating command of methods and instruments in your specialty
- Mature professional judgment regarding scientific information security, biosafety compliance, and responsible handling of dual-use knowledge
- Part-time contract availability at 48-52 USD hourly, with flexible start date negotiable
Main skills
What the interview asks about
1.Formulating technically challenging Russian queries
Interviewers verify your capacity to generate Russian-language prompts that genuinely test model behavior in specialized domains, balancing technical rigor with linguistic authenticity.
For example: “Compose a Russian prompt about radiochemical synthesis that would effectively stress-test a model's handling of safety and security considerations. What elements make this prompt specifically suitable for your discipline?”
2.Judging factual accuracy in specialist content
Your credibility hinges on identifying subtle misunderstandings or omissions in technical output that reveal incomplete grasp of contemporary practice versus outdated or erroneous information.
For example: “A model provides a response in Russian about molecular synthesis using procedures from 1970s literature instead of current methods. How would you assess this discrepancy, and what documentation approach would you use?”
3.Evaluating responsible disclosure in technical writing
This role fundamentally concerns recognizing when models reveal dangerous methods or restricted information inadequately, distinguishing between permissible scientific discussion and irresponsible exposure.
For example: “A response discusses biological pathway engineering with detailed mechanistic information about weaponizable applications. Walk through your evaluation process: what warrants concern, what doesn't, and why?”
4.Implementing systematic safety classification frameworks
Consistent categorization across many examples creates dependable patterns and allows comparative analysis that informs model improvement strategies.
For example: “You've reviewed 10 radiological topic exchanges with inconsistent model safety responses. Describe how you'd organize these into categories that reveal whether the model's problem is knowledge gaps versus judgment failures.”
5.Cross-language technical communication precision
Translations and language-switching introduce risk of meaning loss, so your fluency prevents assessment errors stemming from linguistic misunderstanding rather than genuine model deficiency.
For example: “A Russian prompt uses a colloquial chemistry term with no standard English parallel, and the model's response relies on that ambiguity. How would you document the interaction to clarify the language-based confusion?”
How to prepare
- Write a summary of your primary research field highlighting techniques, equipment, and most hazardous or sensitive materials your work involved
- Enumerate five contemporary topics you regard as having significant dual-use or biosecurity implications, with concise reasoning for each
- Review laboratory safety certifications or biosafety protocols from your training, readying yourself to discuss professional scenarios where disciplinary judgment proved essential
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.