$ focus:
Applied mathematician finishing an MS in Data Science at Pace University (December 2026). My projects cover fraud detection, churn prediction, A/B testing, SQL analytics and retrieval-augmented generation, each with code, held-out evaluation and a written recommendation.
Each one states a question, tests the answer on data the model never saw, and ends with what to do about it.
Compares logistic regression, XGBoost and an unsupervised Isolation Forest on 284,807 transactions, scored on a later time period. Sets the alert threshold by dollar cost.
Predicts which of 7,043 telecom customers will leave, explains each prediction with SHAP and includes a Streamlit app that suggests retention actions.
Analyzes a 90,189-player experiment with a sample-ratio check, z-tests, bootstrap, a Bayesian model and power analysis.
Answers questions about the scikit-learn User Guide with cited sources. Compares keyword, embedding and hybrid retrieval on 50 labeled questions, with a tested Python package and a Streamlit app.
Classifies 50,000 movie reviews and compares a lexicon baseline with Naive Bayes and TF-IDF linear models, then studies where the model fails.
Eight business questions answered in SQL with CTEs and window functions, including RFM customer segmentation and a Pareto analysis of artists.
305K service requests modeled in Neo4j to compare agencies and resolution times.
→ NYC Electricity ForecastingProphet forecasts at daily, monthly and yearly scales, with a borough comparison.
→ Diabetes ProgressionLinear regression vs XGBoost with held-out evaluation and learning curves.
→ Discretization StudyDecisionTable and MDL built from scratch; shows a 3-point accuracy inflation from data leakage.
→ Data Mining TechniquesClassification, regression, clustering and Apriori, with my own OneR, Winnow and PRISM.
→ NYC Urban Life (SQL)Arrests, restaurant inspections and street trees compared across boroughs.
→ Retail Sales EDASales patterns and recommendations from a year of transactions.
→ Parkinson's Telemonitoring EDAVoice measurements related to clinical severity scores.
→What I use in the projects above.
Mathematics first, then software delivery, then data science.
New York · GPA 3.89
Remote, part-time. Planned and delivered software projects with engineering teams.
Batch topper and silver medalist
Taught mathematics.
Open to Data Scientist, Data Analyst, ML Engineer and risk or fraud analytics roles in New York or remote.