Credit Decisioning Infrastructure and Machine Learning Ops

Technical Lead on various feature engineering and data production projects in the Card Data Science Team, owning and leading the R&D and the deployment of critical data solutions for both real-time and batch use cases.- Innovatively developed a distributed ETL pipeline to process data for the full Card Customer Management volume (~100M), partnered with an Engineering Team to successfully deploy into production, and published a Credit Bureau data product for critical business consumption, empowering business to respond to the COVID-driven underwriting changes and re-enter market.- Owned and led a Data Science effort to design features and implement a core functionality, in a multi-divisional collaboration with Business, Product, Engineering teams to establish a complex system for Income Verification. Managed to deliver a robust and reliable solution that enabled a triumphant launch of the new capability.- Recipient of multiple Leadership-nominated highly selective rewards as individual contribution recognition.

Primary designer and developer of feature creation applications that serve Card risk models and credit-related operations in production.- Designed and developed a highly performant application as a Microservice to generate over 1800 Consumer Credit features in real-time for underwriting needs, with performance exceeding platform requirement.- Performed complex analyses on credit data and delivered key results timely in a multi-department collaboration effort in response to a high-profile risk event, leading to an eventual successful remediation.

Modeled, implemented, validated, delivered, and quality monitored the satellite Sea Surface Temperature (SST) data products at the Center for Satellite Application and Research (STAR), NOAA. ● Designed and implemented a new regression model for SST using Regularized Least Squares to considerably boost the robustness of the model and consequently improve the accuracy and stability of the SST products.● Successfully developed an analysis system (3S) with SQL/FORTRAN/C++/R to monitor satellite sensor performance by producing time series of orbital statistics on a near real-time basis.● Expertly designed and maintained a web application to interactively display 3S data results.● Applied extensive statistical knowledge and visualization skills to lead the development of the latest versions of three other web-based systems (SQUAM, MICROS, and ARMS) to validate and monitor both in-house and external enterprise SST products for regional and global coverage. The tools are widely recognized and utilized by research institutions and government agencies from over 30 countries.● Effectively performed various temporal and spatial analyses on geospatial satellite data, published work on peer-reviewed journals, and presented at domestic and international conferences

Performed large scale simulations for 1D bosonic lattice systems with exact numerical approach.● Designed highly parallel algorithms in C/C++ and scaled computation to up to 1200 cores in High-Performance Computing (HPC) environment.● Analyzed massive data by performing coarse graining and identified phase-dependent statistical distributions.● Demonstrated important dynamical response using signal processing techniques such as Fourier transformation.

Studied nonequilibrium properties in quench dynamics through single-particle and many-body analyses.● Built up an efficient and highly parallel program based on large matrix diagonalization using the Lanczos method.● Optimized algorithm for sparse matrix operations by using data structures such as linked list.● Developed codes in C++ and Bash/AWK to process time fluctuation data in gigabytes for bin analysis.

Conducted a project on quantum phase transition in collaboration with a group of senior scientists.● Originally developed a numerical method to compute quantum noise correlations far more efficiently than prior art.● Analyzed and visualized data with MatLab, etc, discovering a new signature of the phase transition.● Organized periodic teleconferences to deliver results to and discuss with collaborators.