Machine Learning

My machine learning projects focus on computer vision applications and implementing core algorithms from scratch. I'm particularly interested in automation and predictive modeling.

A naive bayes classifier designed to detect voicemail beeps in real-time during phone calls. This machine learning project uses audio signal processing and probabilistic modeling to identify the distinctive beep patterns that indicate when a voicemail system is ready to record a message. The system can be integrated into call processing applications to automatically detect when a human has stopped speaking and a voicemail system has taken over.

A comprehensive, reproducible study comparing five optimizers (SGD, Momentum, Adam, AdamW, and AdaBound) on MNIST dataset across different network depths and activation functions. The research includes a complete Jupyter notebook for reproduction and a detailed PDF paper. Key findings show that adaptive optimizers outperform SGD and Momentum for deeper networks, with AdaBound + Swish achieving 98.2% accuracy on 3-layer networks. The study demonstrates the importance of optimizer selection and activation function choice in neural network training.

This repository contains a collection of Python scripts designed to automate different tests available on the Human Benchmark website. These scripts leverage various Python libraries to interact with the user's screen, analyze visual data, and automate mouse and keyboard actions.

This repository offers a streamlined Python decision tree classifier, capable of handling CSV datasets for varied predictive modeling tasks. Utilizing pandas for data management and numpy for calculations, it features an adjustable-depth decision tree algorithm and k-fold cross-validation for optimized performance evaluation.

Software Engineering

My software engineering projects range from mobile applications and browser extensions to web development and API design. I focus on creating practical solutions that enhance user productivity and experience.

This website was created as a team project for CS 373 in Spring 2024. It aims to support individuals facing homelessness in Texas. Our mission is to offer an online platform that connects those in need with homeless shelters, promotes awareness for organizations aiding homeless communities, and spotlights local volunteer opportunities. Additionally, we developed an API to enhance access to resources and information, streamlining the process of finding help and engaging with community support efforts.

Frustrated by the constant clutter of tabs while working on various projects, I embarked on a mission to find a solution. The result of my efforts was Tab Time Tracker, a Google Chrome extension crafted with HTML, CSS, and JavaScript. The idea was simple: create a tool that not only identified duplicate tabs but also sorted them based on usage time.

Auto Snooze sprang from my own desire for an alarm that could turn off by itself, essential for my lucid dreaming practices, specifically the Wake Back To Bed (WBTB) method. I couldn't find anything suitable on the App Store, so I created a solution that allows users to customize their alarm sound and duration.

Game Development

My passion for game development spans from mobile games to immersive 3D experiences. I focus on creating engaging gameplay mechanics and memorable user experiences.

A real-time multiplayer arena shooter game built with Node.js and Socket.io, featuring dynamic weapon systems, level progression, and competitive gameplay. Players engage in fast-paced combat with upgradable weapons and passive abilities, while competing for top positions on the live leaderboard. The game includes a sophisticated upgrade system with legendary status unlocks, real-time player statistics, and smooth client-server synchronization. Built with modern web technologies including HTML5 Canvas for rendering and Vite for optimized development workflow.

Stomach is a first-person horror cooking game where you play as a line cook recently hired for a cruise ship. Your job is to prepare ingredients and make dishes to deliver to customers, but each night the ship seems to twist and turn against you, with walls that sound less like creaking and more like breathing. No matter what you might see or hear... YOU. MUST. COOK. This capstone project will be released on Steam and itch.io.

A multiplayer prop hunt game developed for Highrise Studio where players can transform into props and hide from each other. Built with Unity using Lua scripting, this project demonstrates my skills in multiplayer game development, networking, and creative gameplay mechanics. Players take on the role of either hunters or props, with props able to disguise themselves as various objects in the environment while hunters search for them.

BubPop Supreme was my first large-scale solo project, a relaxing bubble popping game that I launched on the App Store in 2023. I focused on making every pop feel satisfying and built a visually appealing world around it. The game is packed with cool power-ups and serene backdrops, making it the perfect go-to for anyone looking to relax. It also incorporates advertisements and in-app purchases. Check it out! Its free to download.