What the work actually looks like

You write prompts in Thai — many of them deliberately awkward, indirect, or adversarial — across topics the model is supposed to handle carefully: medical and legal advice, self-harm, extremism, fraud, weapons, sexual content, political sensitivities specific to Thailand. You then read model outputs and conversations and classify them against a written rubric that the platform supplies and revises as the project runs. Every classification is accompanied by a short written rationale in English explaining what the prompt was testing, how the model responded, and why the label fits. That rationale is the deliverable reviewers read most closely; a correct label with thin reasoning is often treated as a miss.

A meaningful part of the job is cultural rather than linguistic. Phrasing that reads as benign in a literal English gloss can carry a very different charge in Thai — register shifts, indirect requests, royal and religious sensitivities, regional slang, romanised Thai and code-switching all matter. You are expected to notice when a model responds safely in form but misreads the cultural stakes, and to describe that gap in terms a non-Thai-speaking reviewer can act on.

What the screen looks for

Mercor's process is largely AI-led: a resume parse, then a recorded conversational interview, sometimes followed by a short written or Thai-language task. The interview probes fluency in both directions — expect to be asked things in Thai and to explain your reasoning in English — and pushes on judgment with follow-ups. Screeners are checking whether you can hold a consistent line on a sensitive category rather than deciding case by case, whether you can articulate a dual-use distinction without either refusing everything or waving everything through, and whether your stated availability is real. Vague answers get a second and third follow-up; specifics survive them.

Logistics

  • Fully remote and asynchronous; work is claimed from a queue rather than scheduled.
  • Roughly 7 hours per week, part-time, with an immediate start.
  • Observed pay band of $18–22/hr — rates vary by project and are not guaranteed.
  • Based in Thailand or wider Southeast Asia is a stated preference, not a gate.
  • Expect exposure to distressing subject matter as a routine part of the task set.