Governed Streaming Lakehouse for Patient Vitals Monitoring

Hospital wards increasingly stream patient vitals, heart rate, oxygen saturation, blood pressure, from bedside IoT devices, but a raw firehose of readings helps no one: clinicians need validated, trend-aware views, and hospitals need airtight governance over who accesses medical data. Building that refinery for physiological data is a demanding, realistic data engineering challenge. The intern develops a real-time pipeline on Databricks using Spark Structured Streaming and Delta Lake. Device readings stream in from Kafka into Bronze tables, Delta Live Tables enforce schemas and data quality rules that catch malformed or physiologically impossible values, and the Medallion Architecture refines data through Silver cleaning into Gold aggregates per patient and ward. Databricks SQL dashboards visualize vital-sign trends and flag anomalies such as sustained deviations from baseline, giving clinical staff a live monitoring surface. Governance is treated as a first-class requirement: Unity Catalog controls data access, secret scopes protect credentials, and role-based permissions separate engineering from clinical viewers. The project demonstrates streaming pipeline engineering under data quality and privacy constraints, exactly the combination healthcare employers struggle to hire for, and shows the intern can build analytics infrastructure worthy of safety-critical data.