Velvet provides state of the art multimodal datasets and evals to frontier AI labs. Our team is comprised of former ML researchers who worked at Meta FAIR and Adobe, and we are solving important challenges in the procurement, evaluation, and enhancement of multimodal data.

Our culture is very scrappy, we work long hours and wear multiple hats. If you are looking for a one-dimensional job that involves sitting behind a desk all day and doing one task, we are not a good fit.

However, if you are a fan of the forward deployed model and are willing to travel, work on ops-heavy projects, and help train state of the art models, we may be the right company for you.