Build and maintain batch and streaming pipelines, own orchestration and warehouse loading, model data with dbt, define canonical metrics and semantic layer, ensure data quality/contracts and reliability, and ground LLM outputs in auditable healthcare data.