Cindy Kang Email & Phone Number

San Francisco Bay Area

San Francisco Bay Area

San Francisco Bay Area

San Francisco, California

Stanford, California

Investment Science (Fall 2019)Optimization (Winter 2020)

Foster City, Ca

• Assisted with model validation reports by parsing and cleaning large sets of raw transactional-level data, creating plots and visualizations for exploratory data analyses, and conducting sensitivity analyses of champion/challenger models using R/Python• Migrated Country Risk Assessments (CRA) program from analog to digital by consolidating information from 20+ MS PowerPoints to build out an initial data brick; enriching data brick with country information across risk, finance, and government engagement teams; building out first operational Tableau dashboard; and identifying input gaps, process improvements, and path forward to best-in-class insights on country risks

San Francisco, Ca

San Francisco, Ca

• Coded adjusted roll rate model in R to automate 3-month loss projections for Lendify trusts based on moving average of flow rates, and promptly delivered forecasts to capital markets team each month.• Aggregated new DolEx attributes for use in benchmark model development, finalized model with 15% KS improvement from previous DolEx model. Provided specifications for model implementation and created retro score file for strategy development.• Improved analytics data infrastructure by parsing credit bureau reports from JSON file format to queryable table format. Created analytics data mart for risk reporting/analyses.• Built various regression models on retro data to predict expected charge-off rates on loan-level basis using early performance metrics. Validated model outputs using vintage curves. Improved losses with minimal impact on approval rates by identifying 12 high-risk nodes (out of 9000+) using model outputs.• Monitored customer profile and risk performance for digital marketing channel loans. Used machine learning algorithms to propose policy recommendations which would reduce occurrence of first payment defaults by over 15% and, in relation to previously imposed "quick-fix" rule, decrease approval rate impact by nearly 75%.

Berkeley, Ca

Computing with Data (Summer 2016, Fall 2016)Time Series Analysis (Spring 2017)

Berkeley, Ca

Computing with Data in R (Fall 2016)

Berkeley, Ca

STAT 134: Probability (Adjunct Course)

Berkeley, Ca

San Francisco, Ca

• Completed affinity analysis project that involved data collection/cleaning, variableaggregation, and identification/evaluation of customer affinity on check cashing partner’stransaction data• Improved existing loan application screening process and mitigated losses by identifyingan affinity group with 30% lower first payment default rate than non-affinity counterpart

Ms, Management Science And Engineering, Gpa: 4.16

Ba, Statistics, Gpa: 3.97, Summa Cum Laude