English and Chinese speakers summarise and verify content, solve data and logic questions the model gets wrong, and write explanations that improve a large language model. 40 hours a week on contract, US Pacific overlap. No rate published.
Pay not statedWorldwide