• Own Clinically Grounded AI Evaluation Delivery (Core) • Lead evaluation projects from kickoff → rubric → execution → QA → final reporting • Define milestones, timelines, and delivery SLAs • Manage delivery risk in sensitive, high-stakes domains • Ensure outputs are rigorous, clinically appropriate, and customer-ready • Evaluate safety, appropriateness, and response quality—not clinical outcomes • Familiarity with scope, limits, and responsible handling of sensitive content • Design Rubrics & Annotation Protocols • Partner with mpathic behavioral experts to define rating criteria • Build labeling guides for psychological state, intent, appropriateness, and harm risk • Ensure rubrics balance clinical nuance with operational usability • Drive Quality & Calibration Systems • Implement review tiers: peer review → QA → escalation • Run inter-rater reliability and disagreement reduction workflows • Maintain gold sets and drift monitoring over time • Tooling & Workflow Ownership • Translate evaluation needs into requirements for internal tooling: Work queues and labeling interfaces • Audit trails and version control • Taxonomies for stress, crisis, and safety response categories • Reporting automation and reproducibility pipelines • Metrics & Operational Excellence • Own evaluation program dashboards including: Throughput per expert per day, Agreement and defect rates, Cycle time and on-time delivery, Safety actionability metrics (e.g., customer acceptance, mitigation impact) • Customer & Stakeholder Execution Support • Help scope clinically sensitive evaluation engagements • Translate customer priorities into clear execution plans • Present findings with clinical and operational clarity