JPMorgan Chase & Co.
CCB Risk Modeling Summer Associate, Data Analytics & AI Research
Jun – Aug ‘26 · Wilmington, DE
- Developed PRIMO, an account-level machine-learning framework for predicting highly zero-inflated, right-skewed revolving balances using nearly 10M observations and 2K+ engineered features, supporting downstream interest-income estimation and credit-line decisioning.
- Built an Optuna-based Bayesian feature-selection pipeline with Tweedie XGBoost, gain and SHAP importance, and correlation filtering, reducing the feature space from 2K+ to 100 features.
- Developed nested XGBoost models spanning a squared-error baseline and distribution-aware Tweedie and hurdle-Gamma objectives; tuned each model family separately with 30-trial Bayesian hyperparameter optimization and implemented corresponding PyTorch feature-token Transformers.
- Conducted multi-vintage out-of-time validation across overall and segment-level metrics; the best model reduced interest-income estimation RMSE by 1.4% in Year 1 and 1.9% in Year 2 relative to the existing benchmark.
Department of Statistics, The Ohio State University
Graduate Research Associate
May – Aug ‘25 · Columbus, OH
- Designed DP-GRAMS, a differentially private mode-seeking algorithm for estimating density modes under formal privacy guarantees.
- Analyzed convergence and statistical error for private nonparametric mode recovery.
- Implemented privacy-aware simulation and real-data evaluation pipelines for modal learning, modal regression, and clustering.
View Paper
Statistical Consulting Service, The Ohio State University
Graduate Student Consultant
May – Jul ‘23 · Columbus, OH
- Consulted on interdisciplinary research projects for clients in computer science, speech and hearing science, geography, health and rehabilitation sciences, and anthropology.
- Cleaned data, engineered features, selected models, and translated statistical findings for applied research clients.
The R Project for Statistical Computing, Google Summer of Code 2021
Open Source Developer
May – Aug ‘21 · Remote
- Project: Improvements to
nls(); mentors: Dr. John C. Nash and Dr. Heather Turner. - Developed nlsCompare, an R package for comparing nonlinear least-squares fits through gradient, convergence, and residual-sum-of-squares diagnostics.
- Wrote tests for nlsj, an interim package for improving functionality in base R’s
nls() workflow. - Work led to a publication in The R Journal (2023).
View Project View Paper
Accenture Solutions Pvt. Ltd.
Data Science Analyst Intern
May – Jul ‘21 · Gurugram, India
- Project: Smart Decision Making Using AutoML.
- Built a Python and Dash AutoML GUI for preprocessing, model selection, hyperparameter tuning, diagnostics, and prediction.
- Supported classification, regression, clustering, and anomaly-detection workflows for use cases such as churn prediction, segmentation, and fraud detection.
- Received a pre-placement offer for full-time employment based on internship performance.
