What the work actually involves

You write expert-level prompts in Thai across your scientific subfield — the kind of question a working researcher would ask, not a textbook exercise — and then evaluate what the model returns. Grading covers scientific accuracy, helpfulness, and whether the model handled sensitive or dual-use material appropriately. A third strand is classification: applying a written rubric to sort prompts and conversations into categories, consistently, across many items. Much of the value you add is linguistic as well as scientific: models are weaker in Thai than in English, and errors that would be obvious in an English response often survive translation into confident-sounding Thai.

Dual-use judgment is central rather than peripheral. Chemical, biological, and radiological/nuclear subject areas sit at the boundary where a model should be helpful to a legitimate scientist and unhelpful to someone seeking uplift toward harm. You will be asked to reason about where that line falls in specific cases, and to write down why — not to apply a blanket refusal instinct. No AI or ML background is required; training on the workflow is provided.

What the screen looks for

  • Verifiable doctoral-level depth in one subfield, probed with follow-ups. Breadth claimed across all of chemistry and biology tends to collapse under a second question.
  • Genuine Thai scientific register — can you render technical terminology naturally in Thai, including where Thai borrows English terms and where it does not?
  • Calibrated safety reasoning: the ability to distinguish a mechanistically detailed answer that is standard graduate content from one that provides operational uplift.
  • Rubric discipline — whether you can follow someone else's grading definition rather than substituting your own standards.

Logistics

Fully remote and asynchronous, with an immediate start and a part-time commitment of roughly seven hours per week. Being based in Thailand or wider Southeast Asia is a stated preference, not a gate; applicants elsewhere are explicitly welcome. Pay of $24–28/hr reflects rates observed on comparable Mercor projects and is not a guarantee — banding and task mix can shift by project phase.