iMerit, the annotation firm ExlService bought in July, has opened a platform for recruiting and paying experts who evaluate models rather than tag training data.
The existing community covers more than 30,000 vetted professionals across 118 countries and 106 languages, with backgrounds in medicine, mathematics, agriculture and autonomous systems. The company expects the network to reach hundreds of thousands.
Work runs on Ango Hub, iMerit’s annotation and fine-tuning environment. Recruiting, onboarding, assignment and payment sit in one place, with payouts handled through Trolley.
Tormenting the model
Radha Basu, who heads iMerit at EXL, describes the job bluntly. Specialists question a model and try to break it, catching errors that automated evaluation misses.
Qualification runs in three stages. General screening tests cognitive skill, a second round checks domain depth and how well a candidate works with models, and peers review the most specialized applicants.
Unlike marketplaces that simply match contractors to clients, iMerit keeps project management, quality assurance and auditing in house.
Where the demand comes from
The company argues AI development has shifted from building foundation models to fitting them to regulated work in healthcare, banking and insurance, where a wrong answer carries consequences.
That shift creates steady demand for people who can judge output in a field rather than score it against a rubric, a service cheaper crowd platforms struggle to supply.
