- Machine learning models for lead merchant assignment, marketing campaign uplift model and customer LTV estimation - R&D of new statistical methodologies for Groupon A/B Experimentation Platform- Designing automated data ingestion and testing pipelines

Uptake's predictive analytics software turns data into actionable insights. Our products improve productivity, safety, and reliability for our customers across major industries, such as rail, wind, and fleet. • Core member of Fleet industry vertical data science team, creating machine learning models to predict mechanical failures across different fleet systems• Collaborating with software, database engineers on our capability to deploy data science models at scale into our platform and products• Developing internal data science software and tools for machine learning and analytics • Mentoring social sector data professionals through Uptake.org's Data Fellows Program

 Led a team involved in managing and analyzing multiple, complex datasets, ensuring dataintegrity, quality, and timeliness of results delivery. Creation and maintenance of automated data reporting dashboard system using R markdownfor medical school client. Improved workow efficiency and reproducibility. Trained and mentored new statistical analysts and team members. Advised Ph.D. candidates as a member of graduate thesis committee.ults.

 Providing statistical and database design consultation to clients performing medical research. Responsible for statistical programming, analysis, results interpretation, and manuscript development. Co-authored 15+ publications from these collaborations. Developed R Shiny App for co-variate adjusted randomization in clinical trials. Created supervised learning models to predict kidney disease after liver transplantation frombiomarker data using Random Forest.

 Developed a novel statistical methods to identify dierentially expressed transcripts for high di-mensional RNA-sequencing data. Improved classication sensitivity and specicity in simulateddataset vs traditional methods. Identied risk factors using regression method for electrographic seizures in pediatric ICU pa-tients and developed a prediction model for seizure occurrence.

 Conducted eld study in China investigating how early health risk factors, early nutrition andenvironmental toxicity influence emotional and behavioral development in children. Designed and validated surveys used in the study. Applied statistical methods e.g., GLM, GEEand mixed models to analyze longitudinal outcomes. Trained research sta on survey administration and longitudinal database maintenance.