• Translate medical policy into executable logic - Read and interpret medical policies and clinical criteria (e.g., lab thresholds, temporal windows, trend logic, exclusions). • Convert requirements into correct, maintainable SQL and Python implementations (e.g., creatinine-based AKI rules, bilirubin thresholds, troponin dynamics, ABG-derived criteria). • Design rule representations that are composable and auditable (clear inputs, outputs, assumptions, edge cases). • Prompt engineering and system parameter tuning for AI configuration that extracts clinical information from medical records. • Build robust clinical feature pipelines. • Create and maintain pipelines that compute clinical features from extracted signals (labs, vitals, flowsheets, notes-derived facts). • Handle tricky realities: missing timestamps, multiple measurement sources, unit normalization, deduplication, conflicting values, provenance tracking. • Own measurement, evaluation, and continuous quality improvement - Define and instrument accuracy metrics for the AI system that extracts data from medical records. • Build gold datasets, sampling strategies, and review workflows with clinical/operations partners. • Perform error analysis, identify root causes (retrieval failures, OCR issues, extraction ambiguity, policy interpretation gaps), and drive improvements. • Establish engineering frameworks and tooling - Create reusable tooling for policy-to-code translation: templates, test harnesses, validation suites, regression checks, and monitoring dashboards. • Improve infrastructure for large-scale runs: orchestration, logging, lineage, versioning, and reproducibility. • Implement guardrails and QA gates so policy logic changes are safe, traceable, and measurable. • Partner deeply with domain experts - Work with clinicians, policy specialists, and operations to clarify ambiguous requirements and ensure implementations reflect real-world intent. • Produce clear documentation that explains what the code is doing and why, with examples and edge-case handling.