Mathematics · Data · Software
Math major. I write pipelines, train models, and ship production apps — sometimes all three in the same project.
About
Mathematics major at Austin Peay State University, graduating Fall 2026 with a 3.9 GPA. My concentration is Data Science, which in practice means I spend a lot of time asking why models behave the way they do.
Somewhere along the way I started building the tools too — a property management app that replaced a manual billing workflow, an n8n pipeline that feeds LLM output into Discord in real time, a crepe business website with a full order backend. The math and the building feed each other.
I also work as a Thai–English interpreter and run a food business on the side. Both keep me honest about communicating clearly to people who don't care about your methodology.
Featured
IBM Telco Dataset · Python · Scikit-learn · Power BI
Full consulting engagement on the IBM Telco dataset. Churn models, KPI framework, 6-month forecasts, risk scoring, and an investment scenario analysis — output formatted for a board presentation.
Three classification models benchmarked head-to-head (best AUC: 0.766). A KPI scorecard spanning 11 metrics across customer, financial, and operational dimensions. Six-month time-series forecasts using Holt's Exponential Smoothing. Three investment scenarios modelled with full ROI breakdown — the AI chatbot option came in at 142% ROI, which became the recommendation.
All Projects
Full-Stack · Data Management · JavaScript
Built for a real 140-unit apartment building in Thailand. Handles tenant records, meter readings, billing, and payment slip verification. Features AI batch photo analysis that reads meter numbers automatically, auto-calculates bills, detects fake payment slips, tracks mail, and prints all 140 bills in one click.
Pipeline Engineering · LLM · Workflow Automation
A four-node workflow in n8n: Google Drive change triggers a search, the retrieved document goes through an LLM, and the output lands in a Discord channel via webhook — no manual steps. Good for seeing how LLM outputs behave when wired into something real.
Statistics · R
Wage determinants in R. Log-transformation, dummy encoding, VIF checks to catch collinearity, and residual diagnostics. ggplot2 for the visuals.
Full-stack business website for a home-based Thai/Japanese crepe business — built with React, Netlify Forms for order capture, Google Apps Script webhook to Sheets, and a password-protected admin dashboard. Includes a live ordering system, farmers market assets, and LINE notification integration.
View Live Site →EDA · Python · R · ggplot2
FAA incident data. Built features across time, category, and geography to find patterns that aren't obvious in the raw records — seasonality, operator clusters, phase-of-flight risk.
NLP · Python · Scikit-learn
TF-IDF vectorization, two classifiers compared, precision/recall documented. The interesting part was understanding where each model broke — not just which number was higher.
Reinforcement Learning · Python
Q-learning agent that finds its way out of a maze. The reward shaping is where the real decisions happen — too generous and it stops exploring, too harsh and it never converges.
iOS · Swift · OpenAI API
iOS app for logging BJJ training sessions. You talk through what happened on the mat; the app transcribes it, passes it to the OpenAI API, and writes a structured summary. Built in SwiftUI from scratch.
Chatbot · Python · NLP
Python chatbot that answers type matchup questions and predicts battle outcomes. Intent classification under the hood, Pokémon trivia as the test domain.
Technical Skills
Education
Contact
Looking for data engineering, analytics engineering, or data science roles — graduating Fall 2026. If you want to see code before a conversation, GitHub is the right place to start.