What the work involves

You write expert-level prompts in Hindi across your scientific specialty, then evaluate what the model returns — checking scientific accuracy, usefulness to a technically competent reader, and whether the response handles sensitive or dual-use material responsibly. A second stream of work applies structured classification guidelines to prompts and conversations: deciding which risk bucket a given exchange falls into, and documenting why. Much of the difficulty is linguistic as much as scientific — Hindi technical registers vary widely, and part of the job is deciding when to use Devanagari terminology, when a transliterated English term is what a real practitioner would write, and flagging where the model's Hindi output is fluent but scientifically wrong.

No AI/ML background is required; the workflow is taught. What is not taught is the science. Prompts are expected to reflect how someone actually working in organic synthesis, virology, microbiology, radiochemistry, or radiation physics would frame a hard question — not textbook recall.

What the platform screens for

Mercor runs an AI-led interview before any human sees your file. Expect it to test three things: that your PhD work is real and specific (it will ask about instruments, protocols, failure modes, and follow up on vague answers), that your Hindi is genuinely native-level in a technical register rather than conversational, and that your safety judgment is calibrated — neither refusing every sensitive question nor treating hazard information as ordinary content. Candidates who describe their research at press-release level tend to fail the depth probes.

  • Depth in one subfield reads better than shallow coverage of three
  • Prior grading, peer review, TA work, or red-teaming is a genuine differentiator
  • Coursework or institutional training in biosafety, chemical security, or hazmat handling is directly relevant

Logistics

Fully remote and asynchronous, contractor basis, immediate start. Hours are flexible and largely self-scheduled; most contributors treat this as part-time alongside research, though throughput expectations exist and consistently low weekly volume tends to end engagements. Observed pay for this listing is $23–27/hr, set by the platform and not guaranteed for any individual. India and wider South Asia are preferred for time-zone overlap but explicitly not required.