AI Training Data Manager — Annotation & Quality

Our ML team trains custom models and our biggest bottleneck is data. We have a team of 15 remote annotators and no real quality control process — inter-annotator agreement is inconsistent and we constantly discover bad labels after training. We need a Training Data Manager to own our annotation pipeline end-to-end — guidelines, tooling (Label Studio), quality review, annotator feedback loops, and tracking data quality metrics. This is a hybrid PM/technical role. You don't need to train models, but you need to understand what makes a good training dataset.