Looking for a job

David Joung

@davidjoung

Principal Data Scientist at UnitedHealth Group

At UnitedHealth Group, I migrated pre-pay overpayment scoring for the UnitedHealthcare Payment Integrity program from on-prem Hadoop and SAS to Azure Databricks and Delta Lake. Scoring now lands about six hours earlier, and runtime cost fell about 40%.

I also built an Azure OpenAI (GPT-4) case summarization assist for post-pay investigations, cutting case preparation time about 20% while keeping decisions with human reviewers. I standardized drift and data quality monitoring across production scoring pipelines and owned the team's Responsible AI and model risk management.

Earlier at UnitedHealth Group, I developed an XGBoost ranker that improved hit rate by 15 points in an A/B test. At Cerner Corporation, I built a readmission model for the HealtheIntent platform, and at Arvest Bank I built a recovery propensity scorecard that showed a 5 to 8% lift in recovered dollars in an outreach pilot.

Experience

Work history, roles, and key accomplishments

Current

Principal Data Scientist

Current

UnitedHealth Group

Jul 2021 - Present (5 years 3 months)

Led migration of pre-pay overpayment scoring to Azure Databricks and Delta Lake, improving efficiency and reducing runtime costs. Built an LLM case summarization assist and standardized monitoring across production pipelines.

Azure Databricks Delta Lake Airflow MLFlow Azure OpenAI GPT 4 Great Expectations Power BI XGBoost Model Risk Management Responsible AI PHI HIPAA

Senior Data Scientist

UnitedHealth Group

May 2019 - Jul 2021 (2 years 2 months)

Replaced rules-order audit queues with an XGBoost ranker, improving hit rate and recoveries. Repaired flag lineage and rebuilt anomaly baselines during the 2020 telehealth surge.

XGBoost Hive A B Testing Azure Databricks ETL Anomaly Detection

Data Scientist

Cerner Corporation

Oct 2017 - May 2019 (1 year 7 months)

Built a 30-day readmission model with per-client calibration on a multi-tenant platform. Replaced a legacy risk score and introduced data quality checks for client registries.

Hive scikit learn Python Data Quality Probability Calibration Healthcare Analytics

APS Data Scientist

Arvest Bank

Jan 2017 - Oct 2017 (9 months)

Rebuilt monthly roll-rate and charge-off reporting with parameterized SAS and SQL Server, reducing production time. Built a logistic regression recovery propensity scorecard and ran an A/B outreach pilot.

SAS SQL Server Logistic Regression A B Testing Performance Testing Model Governance

Quantitative Analyst

New Directions Behavioral Health

Oct 2014 - Jan 2017 (2 years 3 months)

Rebuilt discharge outreach lists using authorization dates, reducing staleness. Built SAS readmission flagging and standard reporting for client onboarding.

SQL Server SSRS SAS Reporting Healthcare Analytics