Joon-Ku I.

About Joon-Ku I.

I build machine learning systems that actually ship and get used in production. Over the past 10+ years, I\'ve learned that the difference between a good model and one that creates business value isn\'t just technical elegance—it\'s understanding what problem you\'re solving and making it work despite real-world constraints.At Elevance Health, I\'ve built 40+ production clinical prediction models covering 40 million members. My work includes maternity risk models supporting outreach to pregnant Medicaid members annually—predicting C-section risk, gestational diabetes, NICU needs, and identifying pregnancies early so care managers can ensure OBGYN visits and healthy deliveries. I also developed readmission risk models that help prioritize care management for recently discharged members, plus surgery prediction, cost forecasting, and ER utilization models. These systems have been running in production for 5 years using XGBoost and LightGBM, and I established the development processes and validation frameworks that let our team maintain quality at scale.Recently, I\'ve expanded into GenAI and LLM systems. I built a text-to-SQL platform using PyAutoGen multi-agent framework that lets STARS campaign managers query healthcare data in plain English, plus call center automation tools processing 400K daily calls for real-time summarization and audit extraction. I also lead an 18-member cross-functional team (7 onshore, 11 offshore) and I\'m known as the go-to person for troubleshooting production issues.What drives me is translating ambiguous business requirements into solutions that work—navigating compliance requirements, legacy systems, and enterprise constraints to deliver systems that care managers and business teams actually use. I\'m equally comfortable building traditional ML pipelines or cutting-edge GenAI applications, as long as they solve real problems.Tech stack: Python, XGBoost, LightGBM, PyAutoGen, FastAPI, Snowflake, OpenAI API, LangChainDomains: Healthcare analytics, clinical prediction, cost forecasting, member risk stratificationKeywords: Machine Learning, Production ML Systems, Healthcare AI, Predictive Analytics, GenAI, NLP

Experience

1. Data Science Director, Enterprise Ai

   Anthem Blue Cross and Blue Shield

   Jan 2018 — Present · Chicago, IL, US

   Production Clinical ML Models (XGBoost, LightGBM)- 40+ Models, 40M+ MembersBuilt maternity risk prediction suite for Medicaid populations (C-section, gestational diabetes, NICU risk, early pregnancy ID) supporting care management outreach to 400K pregnant members annually for OBGYN coordination and healthy deliveries. Developed readmission risk, surgery prediction, cost forecasting, and ER utilization models using XGBoost and LightGBM. Established development processes and validation frameworks enabling team to maintain production quality across models running 5+ years.GenAI/LLM SystemsBuilt text-to-SQL system using PyAutoGen multi-agent framework with Snowflake, enabling STARS campaign managers to query healthcare data in natural language. Production pilot with expansion pipeline. Developed call center GenAI solutions processing 400K daily calls for real-time summarization and automated audit extraction. Created FastAPI microservices for LLM deployment.LeadershipLead 18-member cross-functional team (7 onshore, 11 offshore) with 100% production deployment rate. Design causal inference and experimentation frameworks for program evaluation. Primary troubleshooter for production model issues.

Education

• The Ohio State University

  MS, Mathematics

  2003 — 2007

• Northwestern University

  PhD, Industrial Engineering and Management Sciences

  2007 — 2012

• Seoul National University

  BS, Mathematics

  1996 — 2003

Skills

• Credit Risk
• Predictive Modeling
• Dbvisualizer
• Python
• Latex
• Sql
• Ccar
• Spark
• Monte Carlo Simulation
• Sas Programming
• Quantitative Analytics
• Optimization
• Teradata
• Risk Modeling
• Marketing Mix Modeling
• Forecasting
• R
• Multivariate Statistics
• Optimizations
• Mapreduce
• Operations Research
• Hadoop
• Netezza
• Statistical Modeling
• Time Series Analysis
• Statistics
• Econometrics
• Machine Learning
• Credit Cards
• Cecl
• Sas
• Data Analysis
• Analytics
• Data Mining
• Matlab
• Teradata Aster
• Dfast
• Predictive Analytics