Projects
Marketing Strategy Simulation
Problem
Budget was spread evenly across 3 customer segments with no analysis of which actually returned the investment.
Approach
Benchmarked CLV, CAC, and retention across 3 segments covering $727M in expected value.
Tools
Excel (scenario modeling, forecasting)
Impact
Found 14% of accounts drove 69% of expected value ($504M of $727M). Recommended shifting spend to the top-LTV segment.
Fraud Detection & Data Analysis
Problem
Raw transaction data had duplicates and missing values, making anomaly patterns hard to separate from noise.
Approach
Analyzed 1.49M+ records in Python, cleaning and standardizing data from multiple sources before modeling.
Tools
Python (pandas, NumPy, scikit-learn, matplotlib)
Impact
Cut missing-value errors 25% and improved model performance 23%.
Public Transit Performance Analytics
(1st Place)
Problem
Performance reporting was manual and fragmented across 3 fiscal years.
Approach
Wrote SQL across 210K+ records and 60+ routes, then built 4 Tableau dashboards on automated pipelines.
Tools
LucidChart, SQL, Tableau
Impact
Identified 6 issues driving on-time variability and removed manual effort from recurring reporting. 1st place, MSBA database competition.
Warehouse Optimization
Problem
Allocation decisions ran on fixed rules with no way to test better configurations.
Approach
Built two optimization models to evaluate allocation and shipment scenarios.
Tools
AMPL, Gurobi
Impact
Identified configurations increasing projected profit 450%.