Limin Wang, Phd Email & Phone Number

East Brunswick, Nj, Us

Loan Default Detectoro To accurately estimate the default risk of applicants and help loan officers in decision-making, built a Loan Default Detector based on binary classification models.o Processed and analyzed historical loan data, conducted feature engineering, feature selection, and missing data imputation; experimented with multiple classification models, including Logistic Regression, Random Forest, and XGBoost; achieved a ROC-AUC score of 80% with Xgboost.o Wrapped the best-performing model as a Flask App and developed it on AWS; interpreted the default risk of each individual applicant with SHAP values.

136 Frelinghuysen Road Piscataway, Nj 08854-8019 Usa

Predictor of Noval Solar Panel Materialso To accelerate the innovation process of next-generation solar panels with better energy harvesting efficiency and lower the R&D expense, built a Material Candidate Identifier based on binary classification models.o Collected, combined, and processed historical experimental and simulation data from various sources; conducted feature engineering and feature selection based on domain knowledge; experimented with multiple classification models (Logistic Regression, Random Forest, and XGBoost).o Achieved a precision score of 81% with the best performing model (XGBoost), 80% higher than the traditionaltrial-and-error method and reduced the research expense by 55%. Used SHAP values to interpret model outputs and collaborated with cross-functional teams to optimize the material innovation process.

College Park, Md

o Designed and conducted a series of Density Functional Theory (DFT) calculations of superconducting materials. Built an automated pipeline to analyze and visualize the calculation results using Python.

o Participated in the development of Orbital-Symmetry Wannier (OSW), a first-principles calculation software package built with Fortran and C++ for evaluating electronic structures of materials