A Short Summary of My Past Projects

Here's a quick dump of some of the more notable past projects I've worked on, roughly in order of greatest to least significance.

Video Interpolation - Source

A webapp that allows you to upload a video and effectively double its framerate using a custom-built and trained convolutional neural network. The tech stack is a Vite/React SPA, which interacts with a Spring Boot file server, which finally sends the video files to a Python/PyTorch worker process that enqueues and processes the given videos.

Next Frame Prediction

As part of a my introductory machine learning course in Summer of 2024, I implemented a simple CNN that would predict the next frame of a set of frames from a video. It was trained on the UCF101 dataset in PyTorch.

Minecraft Server Stats - Demo, Source

A website made to track the stats of a Minecraft server. Uses result caching to avoid overloading any particular server with requests. Made using React and Spring Boot.

SaraAI - Source

One of my first attempts at utilizing a publicly-accessible REST API, this was a Discord bot which would listen to user messages and track the frequency of words and the frequencies of each of their next words. Periodically, this information was backed up to a JSON file to be reloaded on startup, and either randomly or by a direct command, the bot would randomly choose which words to say until it encounted a sentence ender.

Long Exposure Averager - Source

A command-line application written in Java and OpenCV to read video or image pixel data to create a long exposure image by stacking frames on one another, taking the average of every pixel in the given dataset. A good idea for the future would be to use neural networks to enhance the output, as the current data processing algorithm is deterministic and naive.

UCF's Declassified Guide to Committing Tax Evasion - Source

Made in 24 hours at Hackabull 2023 alongside two other devs, a 2.5D infinite runner game that tracks how much you've ran from the feds. Features jumping, sliding, and three lanes to avoid various obstacles.

Conway's Game of Life - Demo

Made in Godot Engine, implements the algorithm for Conway's Game of Life using hashmaps for arbitrary placement and size. Fun fact: I was inspired to develop this project because I was pretty unsatisfied with the implementation of the GoL that my then-professor of philosophy at UCF was using, so I made this game in a week and gave it to him on a thumbdrive. As far as I can tell, he used it in subsequent semesters as part of his lessons.

This article was updated on October 5, 2025