Inside the AI interview: how we vet experts
How a structured AI interview tests real expertise, why being qualified isn't the same as being good at teaching a model, and what we look for.
Every person in Kodiva’s network, whether a software engineer, a surgeon or a tax specialist, has been through the same first step: a structured AI interview in their own field. Here’s how it works, and why we built it this way.
Why not just check CVs?
A CV tells you where someone has been, not what they can do today. In hiring, that gap costs weeks. In AI training work, it’s worse: the job is to judge and explain answers, and a strong résumé doesn’t guarantee someone can do either.
What the interview tests
The interview adapts to each person’s field and experience. It covers three things:
- Real knowledge: questions and scenarios from their specialty, not trivia. An engineer works through a real problem; a clinician reasons through a case.
- Explaining their thinking: can they say clearly why an answer is right or wrong? This is the core skill for training AI.
- Knowing their limits: the best experts say “I’m not sure” when they aren’t. Confidently wrong answers are the most damaging thing an expert can give a model.
What happens after
People who pass are checked further before paid work: identity verification and an NDA for every expert, plus field-specific checks where needed. Then they’re matched to projects by specialty, country and language. A second expert reviews a sample of the work, so quality is tracked over time, not judged once at the door.
What this means for experts
If you’re an expert, the interview is your way in. There’s no need to have worked in AI before. What matters is knowing your field deeply and explaining it clearly. Projects are remote and flexible, and you can see the rate before you accept any work.
Take the interview and join the network.
What this means for AI teams
It means the experts on your project were tested on the skills your project needs, before the project started. That’s why we can staff specialised work quickly from a network of 40,000+ experts in 60+ countries.