Frequently asked questions

Masters and PhD students and graduates, and working professionals with real expertise in one of our six tracks. You can apply from anywhere; we are starting with talent from India.

Five short multiple-choice questions on judging AI outputs, plus one written question about your field. It takes about 10 minutes. We don't show the correct answers.

A practitioner in your track reviews your background, score and written answer. If you're shortlisted, we email you to book a short interview.

You passed the assessment and a reviewer check in your track. Only verified experts are staffed on client projects.

Pay depends on the project and your track. Rates are agreed with you before you start any work, and you are paid per project.

It's flexible and remote. Projects list their expected hours up front so you can decide whether it fits.

Never. Gnara never charges experts — not to apply, not to get verified, not to work.

We use your application only to evaluate you and match you to projects. We never sell personal data. See our Privacy Policy for details.

These are open roles on established AI training platforms, listed so you have work available while Gnara's own projects open. Some are referral links and Gnara may earn a referral bonus at no cost to you.

Human evaluation, fine-tuning data, rubrics, red teaming, domain QA, Indic language work and custom datasets — in the format your pipeline expects.

We agree the scope, write guidelines and rubrics with you, and run a small pilot batch so you can check quality before scaling.

Only verified experts work on projects. Work goes through multi-layer review and reviewers are calibrated against each other.

Yes. We sign NDAs, limit access to the experts staffed on your project, and handle your data only for the work you ask for.