Build Generative AI boosted data product. Integrate LLM to renovate search and recommendation solutions.Lead the development of deep learning spam lead message detection model that hit over 97% recall and 97% TNR rate; Built end to end feedback loop and streamlit verification interface for automated model updating.

● Lead the Recommendation Model Evaluation team, designed and built recommendation model training and evaluation pipelines.● Conducted research and experiments to improve recommendation results, and e.g. developed DNN based approach to resolve data issue and enhance personalization. ● Designed and built training data quality and drift monitoring system and delivered with Looker dashboards.● Designed and create project roadmaps, managed team and tasks and deliver solutions to stakeholders.

• Research and built pricing models, develop and deploy ML pipeline for item price recommendation.• Managed and Led the development of pricing analysis platform for customer acquisition and B2B contract formulation. (McKesson Global Award Nomination)• Developed and deployed item recommendation API for McKesson eCommerce platform.• Built and deployed Enterprise Anomaly Detection Systems (with ML pipeline, Web Application and API).• Explored the product pricing competition landscape and identified growth opportunities for over 4000 products through exploratory data analysis. Proposed contract restructure solutions and partnered with the product team to launch the campaign. It drove over $20 Million profit growth during 2020.

● Improve the dynamics of Penn State engineering and science graduate community by connecting students from multiple academic disciplines with consulting-related training and career opportunities.● Organized management consulting case-interview workshop.● Invited speaker and coordinated consulting training seminars.● Finalist of Johns Hopkins University 8th Annual International Healthcare Case Competition by providing business development strategy for Next Healthcare Inc. (Top 5)

• Created and managed the Pennsylvania roadway and intersection crash database for PennDOT. Developed the Pennsylvania-specific crash predictive models and improved the prediction accuracy by 2.7 to 37.2 percent compared to the Highway Safety Manual models.• Introduced the Bayesian propensity score matching framework to analyze observational traffic safety data. Improved the accuracy of effect estimates by 9.6 percent and developed the simulation-based sensitivity testing system to evaluate the robustness using R and JAGS. (Dissertation research)• Developed predictive models to estimate the pedestrian and bike traffic at Minneapolis urban intersections. Implemented feature selection algorithms to evaluate the importance of socioeconomic and land-use features. Applied k-fold cross-validation to compare the model performance.• Calibrated settings of the dynamic curve warning system for the U.S. Federal Highway Administration. Defined speed performance indicators, designed traffic experiment, collected data and performed statistical tests to compare the functional scenarios.• Introduced survival models to forecast the pavement friction degradation using field test data. Identified the pavement degradation cycle and provide maintenance strategies for the Pennsylvania Turnpike Commission.• Analyzed the demographic and survey data and estimated the daily trip and mode choice demand in Washington. Identified the impact of shopping preference and personal business on travel behaviors.• Used the SHRP2 Data to develop a database of horizontal curves with guiderail and delineators. Develop the safety performance indicators for FHWA using propensity score matching.•Assessed the benefits of lowering speed limits in speed reduction and safety improvement for Montana DOT. Estimated linear and quantile models, and use empirical Bayes before-after evaluation for speed and safety analysis.

•Launched a pedestrian study to support the terminal crosswalks consolidation. Designed experiment, collected pedestrian crossing data using video cameras. Analyzed the crossing volume and vehicular traffic, and identified the safety and operation benefits. Delivered recommendations to the executive office.•Mitigated airport parking conflictions by reconfiguring ground transportation parking zones. Create work orders for the curbside repainting plan. Edited and updated SFO curbside parking regulations.•Managed and examined temporary traffic control plans for all airport construction projects.•Conducted a survey to investigate the traffic and environmental impact of of Uber/Lyft vehicles on airport road network.