Projects

A full-stack application that allows users to watch YouTube videos together in real time. Features include synchronized playback, video queuing, and room-based collaboration. Built with React on the frontend and a Node.js/Express backend, with WebSockets enabling live synchronization between users.

Stack: React · TypeScript · Node.js · Express · Socket.IO · Tailwind CSS

Personal portfolio and blog built with Next.js, TypeScript, and Tailwind CSS. Statically exported and deployed to GitHub Pages.

Stack: Next.js · TypeScript · Tailwind CSS · MDX

A Discord bot built for a small friend group with a wide range of features, including an LLM-powered chatbot with OpenAI and Anthropic support, guild XP and leveling, Fortnite stats, weather lookup, server theming, and moderation. Built with a database-backed architecture, a pytest test suite, and Docker deployment.

Stack: Python · disnake · PostgreSQL · Docker · OpenAI · Anthropic

Infrastructure-as-code repo that manages personal cloud resources on AWS. Includes CDK stacks for static site hosting, a Lambda-based status monitoring API, a Lambda that auto-starts Minecraft EC2 instances on DNS query, and Docker Compose configuration for self-hosted services.

Stack: Python · AWS Lambda · AWS CDK · Docker · GitHub Actions

A personal status dashboard that monitors self-hosted infrastructure in real time. Tracks Minecraft server status and player counts, HTTP health of web projects, and Docker container resource usage. Built with React and Vite, pulling from a Lambda-backed API on a 60-second interval.

Stack: React · TypeScript · Vite · Tailwind CSS · AWS

A web app that takes any uploaded image and rearranges its pixels sorted by their distance from black, producing a gradient-like visual effect. The sort runs in a Web Worker to avoid blocking the UI during processing.

Stack: React Router · TypeScript · Tailwind CSS · Vite

Collaborated with a team of four to design and implement a convolutional neural network model that classifies Pokémon sprites within battle images, achieving up to 86% accuracy using image processing and deep learning techniques.

Stack: CNN · OpenCV · Python · PyTorch

A method to detect toxic statements, identify specific toxic words, and censor harmful language in online spaces. Uses transformer-based NLP models to flag and redact toxicity at the token level.

Stack: Python · TensorFlow · NLTK · FastText

Predicts NFL game point spreads by testing multiple regression approaches — linear, ridge, lasso, random forest, support vector, and neural network — against team records and per-game metrics.

Stack: Python · PyTorch · scikit-learn · BeautifulSoup