Work and research roles

Built ARIMA/LSTM forecasting pipelines for non-stationary time series and diagnostic workflows for feature stability, forecast-error drift, and signal degradation across regimes.

Tutored undergraduates in calculus, algorithms, and proof oriented problem solving.

Worked on USDA grant-funded applied statistics for heteroskedastic economic time series, including Box-Cox reparameterization, constrained lambda profiling, likelihood comparison, Jacobian correction, and positivity constraints.

Built AI assisted training and retrieval prototypes for course operations.

Built SQL reconciliation, validation, anomaly-detection, and Python/SQL settlement workflows across 1M+ transactional rows, reducing manual reconciliation effort by approximately 25%.

Built Bayesian inference models for network censorship detection and uncertainty quantification.

Led motif census experiments across complex networks for publication support.