What the work involves
You will build and critique psychiatric content in Indonesian and English — clinical vignettes, differential reasoning chains, assessment write-ups, and model-response critiques. A typical task set might ask you to construct a case of a patient presenting with somatic complaints that map onto a culturally specific idiom of distress, then judge whether a model's response handles it as a psychiatric presentation or mistranslates it into a DSM label that doesn't fit. Other tasks are pure evaluation: two model answers to a clinical question, and you rank them and write why, in enough detail that a non-clinician reviewer can follow your reasoning.
The bilingual demand is real and not cosmetic. Much of the value here is in catching where a model's Indonesian output is grammatically fine but clinically or culturally wrong — register that would be inappropriate with a patient's family, translated terminology that no Indonesian clinician uses, or advice that assumes a referral pathway or insurance structure that doesn't exist in Indonesia.
What the screen looks for
- Verifiable credentials: medical degree, completed psychiatry residency, and an active licence. Expect to be asked for specifics on where you trained and where you practise.
- Genuine bilingual production, not passive comprehension — you will likely be asked to explain a clinical concept in both languages or discuss how you'd render a term with no clean equivalent.
- Evaluation judgment: whether you can separate "answer I disagree with" from "answer that is unsafe or wrong," and whether you write critiques that are specific rather than impressionistic.
- Handling of sensitive content: self-harm, psychosis, coercive family dynamics, and religious framing all appear in the material.
Logistics
Fully remote, contractor status, asynchronous. Volume fluctuates with project phases — some weeks offer steady batches, others little. Most contributors treat this as part-time alongside clinical practice, typically 5–15 hours per week. Pay in the $100–200/hr range has been observed on similar micro1 clinical projects; rates vary by task complexity and are set per engagement, not guaranteed.