Nathan Smith

Nathan Smith

Staff Data Scientist @ EarnIn
Dallas, TX · ML · Forecasting · Optimization · Simulation

Data scientist with 6+ years turning messy business problems into measurable impact through machine learning, forecasting, optimization, simulation, and rigorous experimentation.

At EarnIn, I own end-to-end credit risk policy for the flagship product — >$1B in originations and >10M credit decisions monthly. I built a discrete-time simulation framework and a constrained optimization engine to set credit strategy by trading off loss, margin, and LTV, and I drive the experimentation strategy behind policy and model changes.

Earlier, at Republic Finance, I built forecasting and response models (Random Forest → LGBM, 800+ features) that cut mailing volume ~45% while retaining ≥97% of NPV, generating ~$1M/month in savings, and automated reporting pipelines that cut turnaround ~95%.

B.S. in Industrial & Systems Engineering, Virginia Tech.

Python SQL Databricks scikit-learn Optimization Simulation Experimentation

A client-side-only card optimization calculator that considers your spending profile and popular card programs to recommend how to maximize rewards.

Everything runs in the browser — no data leaves your device. You enter your spend across categories, and it evaluates category multipliers and reward structures across well-known cards to surface the optimal setup for your habits.

Client-side Optimization Personal Finance Web App