Assistant Data Science Engineer
Current- Participated in developing Rx Predictive Model API by using Node.js and AWS SDK: fetched data from DynamoDB, generated model features(adherence ratio, drug combo, etc), applied spline transformation on data and scored the data by using spline regression model, and made the API’s median response time less than one second. Also conducted unit, validation, and integration tests and deployed the API on AWS Lambda by using Jenkins- Designed, developed and implemented a Rx1 scoring tool by using Python: used object oriented design to generate severity scores based on rules and prescription data, significantly improved the underwriting process- Participated in Rx Predictive Model batch scoring tool on Spark: designed case classes, partitioned the data, generated drug dosage features and diabetic features, tuned spark configurations and did code refactoring on a billion scale dataset, making the running speed of the program 120 times faster- Wrote customized Dockerfiles for Domino compute environment for different projects