Archive

Applied NLP to 1,000 Amazon reviews for SharkNinja robot vacuums, extracting sentiment and recurring themes to surface actionable product insights. Built while working on robot data at SharkNinja.

Analyzed behavioral data from ~12,000 online sessions to predict purchase intent. Covered the full ML pipeline: feature evaluation, class imbalance handling, and hyperparameter tuning across multiple classifiers.

Predicted whether teacher funding proposals on DonorsChoose.org would be approved, using NLP on project descriptions, word clouds for exploration, and ensemble classifiers for prediction.

Content-based recommendation engine over 12,000+ anime titles using TF-IDF and cosine similarity. Applied K-Means clustering and PCA to explore the latent structure of the dataset.

Compared two forecasting approaches, Seasonal ARIMA and an LSTM neural network, for predicting household electricity consumption, with stationarity testing and full data pipeline.

Modeled admission likelihood from GRE scores, GPA, and research experience using Multiple Linear Regression, Random Forest, and XGBoost. All analysis and visualizations done in R with ggplot2.

Group project analyzing and clustering 20,000 newsgroup posts across 20 categories using NLP, dimensionality reduction with PCA, and visualization. Code not public; report available on request.

Predicted risk of unplanned ICU transfers from clinical data using association rule mining and Gurobi optimization modeling, an early foray into healthcare AI before it became my career.