Partner with ML engineers on feature engineering, dataset versioning, and reproducibility, translating operational and clinical context into the features and training datasets models need, and document architecture, schemas, data contracts, and operational runbooks in Confluence. Negotiate and enforce data contracts with producing teams, establish automated validation across completeness, accuracy, consistency, and timeliness, apply confidence scoring to feeds that business and clinical claims depend on, and detect missing, stale, or implausible records upstream of the models.