Designed and implemented architecture, supporting thousands of API calls daily, to automate prior authorizations reducing decision time by an average of 10 days
Optimized ML training pipeline by decreasing time to process millions of medical documents from 20 hours to 30 minutes by implementing efficient ETL pipelines in Databricks using PySpark
Built a highly scalable NLP feature generation pipeline using FastAPI and Azure Kubernetes Service to be used by other core engineering teams
Developed and maintain internal-use Python packages for data processing, code generation, and feature computation to add major functionality to AI products