Projects

RAINDO is an AI that can assist with technical setup of a randomized video game. Finetuned and attentive to responses from volunteer help professionals, this bot can accurately assist with any questions about setup, logistics, FAQs, and common occurrences. The (discord) bot is also compliant with server rules, flags inappropriate queries, and can respond to questions in 55 supported languages. Increased accessibility for non-English speakers was the main motivation for this project.

Sep 15, 2024

A GPT knowledge file web crawler in base python. Inspired by Builder IO’s GPT Crawler and the Python interpretation by A-baoYang, I decided to create my own version which is a bit lighter on prerequisites. Despite the prior works being very accessible, installing any combination of node, chromium, and playwright can be a pain if your hosting provider does not support installing these dependencies. Therefore, I rewrote the project using only Python’s os, requests, and beautifulsoup4 libraries.

Jul 30, 2024

The Grid Tracker is an upgrade to the supporting window of the Kingdom Hearts 2 Randomizer tracker. Utilizing a game inventory tracking system, the grid updates without user input. The grid comes equipped with UI features such as right-click markdown, custom win-conditions, customizable check menu, color and icon customization, and more.

Apr 14, 2024

Inspired by the Ken Jennings board game Half Truth, I created an online version where leveraging API calls to state of the art AI models to generate true and false statements about any subject. It comes included with color coded UI responses, a scoreboard, and custom game settings. It serves at the ultimate human Turing test and is optimally played when the categories are obscure and generally unknown to all players.

Oct 26, 2023

A spinoff/supporting tool of the project “Bayesian Analysis of Mask Mandate Effectiveness.” I created a minimalistic website with an interactive U.S. map that allowed users to download estimated month-to-month populations of any subset of states in the U.S.. Users could customize the states, the start and end dates, and the interpolation method. The outcome was a downloaded excel file that had an individual sheet for each state with the estimated population each month over the selected range.

May 30, 2022

“Comparison of Efficient Optimizers” was a formal project write-up that compared state-of-the-art optimizers against each other. Convergence and performance were analyzed on two benchmarks to identify which optimizer researchers should use and when.

Apr 26, 2021

Part of the final project for STATS 551: Bayesian Statistics. Leveraging population information from Census data and interpolating populations month-to-month, we built a hierarchical Bayesian model to estimate the effectiveness of mask mandates in all states in reducing Covid-19 cases. tl;dr Mask mandates were effective in reducing cases; primarily so when in response to a surge in deaths in a given month.

Apr 20, 2021