Rohan Rishi

Majority owned portfolio company of Blackstone. ● Analyzed $700M Home Improvement loan pool across CPR/CDR/CNL, UPB trends, FICO drifts, catastrophe exposure, and recovery rates to inform investment decisions; continuing purchase of forward flows agreements. ● Working on model building and clustering analysis of on-going deals to generate the appropriate pricing and analyzing the current pool profile; distribution comparisons and other ad-hoc analysis to support deal structuring of things like rate curves, prepayment and loss curves, FICO distribution, sensitivity analysis etc. ● Led $300M credit card portfolio analysis; applied predictive modeling techniques to loan tape panel data to build forecasting for defaults and drive profit-maximizing strategies. ● Engineered stratification frameworks and dashboards to monitor portfolio KPIs (FICO, vintage, APR), enhancing asset management insights and reporting efficiency.

● Developed processor prediction models using XGBoost, SVC, random forests, autoML; conducted feature engineering & dimensionality reduction using methods like PCA, t-SNE, autoencoders. ● Enhanced model performance using ensemble methods & Bayesian optimization methods (like Tree Parzen Estimators) for hyperparameter tuning; successfully deployed high-velocity production level solution with an impressive ~30ms response time & 96% accuracy. ● Crafted fraud detection models achieving 92% accuracy on chargeback data with bagging models; developed pipeline for processing big data using autoencoders & conducted anomaly detection.

● Alpha research on regime detection & regime-based portfolio optimization; ran Monte Carlo simulations to generate regime dependent price paths; achieved high accuracy scores; created a dashboard. ● Achieved significantly better Sharpe ratios (+3%) based on HMM (Hidden Markov Models) & Wasserstein K-means clustering coupled with mean variance optimization & risk parity portfolio construction.

● Model validation for Probability of Default(PD) and hybrid PD models ● Backend model implementation for major banks (python) ● Extensive use of regression (linear, ridge, lasso, logistic) and assumptions testing for credit models ● Bayesian estimation for scorecard models ● Statistical and Machine Learning models for credit risk models ● Racial and gender bias mitigation techniques in credit models ● Awarded the intellectually stimulating excellence award for being for projects twice

Core courses: Bayesian Statistics, Econometrics, Time Series Analysis, Stochastic Processes, Stochastic Calculus, Deep Learning, Financial Data Science, Financial Risk Management, Fixed Income Markets, Investments & Derivatives, Equity Markets, Currency Markets