• End-to-end model development to support business, including data preparation, model estimation, model implementation, model performance testing, and production deployment• Financial engineering: design and implement software applications addressing the needs of various clients across the firm• Big data analytics (consumer/commercial client data, market data, and financial data), with goals to drive business decision• Business domain: wholesale credit risk, mortgage credit risk, stress testing (CCAR/CECL/IFRS9), capital management, liquidity risk

• Validation (including model framework design assessment, data processing, model estimation, model performance testing) on various statistical models: portfolio loss forecasting model with component models that predict whether a customer would default, loan exposure at default, and collateral value at default; scoring model that predicts whether a customer would become delinquent• Developed a streamlined tool to pull data from model inventory database and generate automated report

1) Solved a fundamental, long-standing question in planetary science and published a well-received paper (editor’s highlight).2) Processed multi-dimensional data sets (~1TB) from observation and numerical simulation in Python/MATLAB, read and wrote specific chunk of data from binary files, analyzed the data with statistical method such as time series analysis, regression and principle component analysis, abstracted the insight of the data and visualized the data in clear graphical representations.3) Creatively proposed a physical explanation for the observed fact from data.4) Developed a model with a PDE system in MATLAB/FORTRAN, successfully provided a quantitative verification for the proposed theory and further extended the theory in more general cases.

1) Established the first mathematical and physical description for a generic type of small-scale instability in shear fluids. 2) Developed a linear instability analysis model in MATLAB/Python to solve large-size, sparse matrix eigenvalue problems. This model is applicable to a variety of linearized PDE systems and therefore has been frequently used by research community.3) Developed a numerical, PDE time-integration model in C++/FORTRAN paralleled with MPI, implemented with a couple types of grids and boundary conditions, and a variety of numerical methods such as finite difference, finite volume, spectral method, forward and backward Euler method and Runge-Kutta method. 4) Processed multi-dimensional large data sets (e.g., NASA satellite data products, global and decadal scale of high-resolution climate data) and analyzed the data with various statistical methods.

TA for undergraduate class AOS 1, led three 1-hour lecture discussions per week, 6 lab experiments/quarter besides grading.

Analyzed global, multi-dimensional atmospheric winds, pressure and temperature data (60 years, ~1T) with statistical methods to reveal the teleconnection between the northern-hemisphere winter storm tracks and the Arctic Oscillation.