Healthcare Analytics Platform
My roleML & Full-Stack Engineer · owned the platform from data pipeline to analytics UI
End-to-end Gen-AI analytics platform delivering $20M+ annual savings for a major pharmacy chain.
Overview
End-to-end analytics platform for pharmaceutical reshipment optimization: an ETL pipeline consolidates millions of unstructured shipment records, a Gen-AI extraction layer turns them into analyzable signals, and an interactive dashboard lets operations teams explore the drivers behind costly reshipments. The analysis surfaced an estimated $20M+ in annual savings.
The Problem
Pharmaceutical reshipments were a major recurring cost, but the reasons behind them were buried in millions of unstructured operational records spread across systems, too messy for traditional analytics to explain why shipments were failing.
The Approach
- 01Engineered an ETL pipeline (PySpark on Databricks) that ingested and normalized millions of unstructured shipment records into a PostgreSQL analytical store
- 02Designed a Gen-AI feature-extraction layer over the consolidated data to isolate and rank the drivers behind failed and repeated deliveries
- 03Built a FastAPI backend and React dashboard so operations teams could query, filter, and visualize those drivers themselves
- 04Translated the analysis into concrete, ranked intervention opportunities rather than static reports
The Outcome
Key Features
- ✓High-volume ETL into an analytical store
- ✓Gen-AI feature extraction over unstructured records
- ✓Interactive driver-analysis dashboard
- ✓$20M+ estimated annual savings