Leading analytical & AI projects to accelerate R&D to Speed Medicines for patients and using innovative technologies & analytical solutions to consistently deliver superior results in high-quality, and services of all supported clinical systems and analytics.1. Trial Design Extraction Program: Design and create the program with NLP algorithm for GSP team that achieves automatically review, extract and generate required information from study protocol files to save 10+ hours per study per protocol, significantly easing the preparation of submission package and enhancing the efficiency of trial design.2. Patient Dropout Risk Management: Led AI research project (POC, use cases, ML solutions) that offers patient risk profiles by ML to predict the risk of patient dropout for next visit in order to improve clinical trial outcomes and efficiency.3. Business Analytics: Support clinical Drug Supply team (CDSIP) to measure the business performance and discover the challenge causes slowing down the supply process by finding data insights and creating analytical metrics (KPI).4. Collaboration & Mentorship: Demonstrate effective communication skills in aligning stakeholder needs with tailored solutions; led technical workshops to share domain expertise, guiding junior data scientists in problem-solving, project design, and storytelling skills to enhance overall team proficiency and quality of solutions.

Research on health claim data, diagnosis procedure codes and medications to develop and maintain state-of-the-art models into products to help patients early health intervention and improve health outcomes in order to mitigate challenges and costs.1. High-Cost Claimant Prediction: Constructed/deployed a Gradient Boost classifier in XGBoost framework to predict the risk of patients being high cost with geography, clinical cohorts, medications and time-series features, which increased 27% performance than vender’s solutions and targeted 1600+ high-risk patients in 6 months.2. Engagement Analytics: Qualified the long-term effectiveness of care management programs by designing the metrics (PMPM, risk) and analytical study (randomization, A/B testing), and quantified $900+ average annual savings of engagement by regression analysis for employers.3. Risk Score Adjustment: Created patient-level risk scores across 6 health service categories to explain individual health expenditures and future cost projections, improved 10% than the market standard by calculating the benchmark and predicting future costs using ML algorithms.4. COVID-19 Dashboard: Analyzed COVID impact for 200+ employers to gauge the safety of reopening business by county-level risk level, time series forecasting, and extracting employees’ attitude by NLP

Collaborated with business teams to develop a self-analytics platform for HCSC (Health Care Service Corporation) that provides automation data pipeline, real-time statistical analysis and interactive dashboards, delivering healthcare data-driven insights: customer segmentation, outcomes transparency, health plans utilization, and chronic condition analysis.1. Predictive Modeling: Predicted health plans utilization, revenue trend and loss ratio for payers by building statistical models (regression and time series models) in R, increased 5% health plan effectiveness.2. Health Plan Analytics: Developed evidence-based health analytics: cost & utilization, HCC risk, diagnosis metrics, customer segmentations and chronic condition, saved stakeholders 95% resources on ad-hoc analysis.3. Deployment: Productized data analyses using batch processing, caching and parallel computing techniques to produce over 30k monthly reports within 45 hours, achieved 75% time reduction and 5x faster than competitions.

Accomplished and presented the consumer segmentation analysis and key differentiators between groups to marketing teams. Categorized 5k+ patients in Texas area into six groups using demographics and healthcare survey to support marketing teams navigating patients differently, which drives business decisions to be better tailor marketing efforts and interact with healthcare stakeholders.1. Consumer Segmentation: Transformed 5k+ consumers demographics and health behavior survey data into 6 segments by K-Means; visualized in-depth evaluation of variable factors and consumer portrait by logistic regression and Tableau.1. Summarized 10 health behavior themes from an internal survey of 5k+ adults, resulted in a six-cluster solution by K-means; conducted segmentation analysis in Tableau to identify key differentiators and discrimination between segments.2. Illustrated scenarios and decision-making processes for future patients by classification models, which boosted efficiency by 28% for healthcare stakeholders to better target, attract, and retrain patients.

Assisted Data Scientists researching on high-frequency trading activity in price competition by processing the daily stock market data (TAQ) and implementing National Best Bid and Offer (NBBO) algorithm using Python and HPC clusters. reproducing discontinuous drops in price competition above $1.00 cutoff.1. Set up high-performance computing (HPC) cluster to perform parallel computation; maintained the GitHub development environment such as documentation, automation process and testing flow.

My major responsibility is that providing data science consulting services helps students and researchers solve challenges from academic projects and papers, covered finance, economics, and social science. I also operated weekly training workshops in D-lab such as organizing students, preparing handouts for workshops, and collecting the feedback from students.https://dlab.berkeley.edu