Portfolio

Projects

Selected work across agentic AI, LLM evaluation, predictive modeling and natural language processing. Every project is written up end to end — the process, the numbers, and what I'd do differently.

X Arcade — Decoy

A Grokathon hackathon build: a multiplayer game where you spot the Grok-written reply among four human ones in thirty seconds.

AI Model Guide

A self-updating catalog of 56 frontier models, with an autonomous research agent that writes new entries and a benchmark suite that scores models on launch day.

Diabetes Risk Predictive Modeling

A logistic regression pipeline over 9,538 clinical records that flags high-risk patients at 97% accuracy and 0.99 AUC.

Task Manager

A Java and PostgreSQL task manager over JDBC — full CRUD, prepared statements, and a schema designed from scratch.

Amazon Purchase Predictor

A five-model bake-off on customer spending, won by a random forest at RMSE 0.113.

LA Airbnb Analysis

NLP topic modeling on guest reviews plus a random forest on Superhost status, shipped as an interactive Tableau dashboard.