CatchBack Cards♠️

CatchBack Cards is a trading card platform that specializes in reducing friction and addressing liquidity issues in the collectibles marketplace. The platform aims to help users liquidate collections at fair value, increase transparency in custom pack systems, provide professional vaulting services, and create an integrated marketplace. CatchBack Cards is built with Next.js 14, TypeScript, Tailwind CSS, and Supabase, and deployed on Vercel.

LLM Plays Battle Tetris👾

My partner and I developed an AI model to play Tetris by fine-tuning a large language model (LLM) using gameplay data from top-ranked Tetrio players worldwide. By extracting key parameters from gameplay videos—such as board state, current and upcoming Tetriminos—we trained the model to predict optimal piece placements and generate corresponding keystrokes. After overcoming challenges in data collection and latency, we built a custom dataset using computer vision and optimized our pipeline for real-time performance. Currently, we are enhancing model efficiency, training on larger datasets, and exploring competitive Tetris strategies such as speedrunning.

House Value Predictor🏠

The House Value Predictor leverages machine learning to provide accurate real estate valuations, addressing limitations in traditional appraisal methods. Using property data from cities across Metro Atlanta, the project compared multiple models including Linear Regression, Ridge Regression, Random Forest, and XGBoost, with XGBoost emerging as the most effective. Through refined feature selection and optimized model performance, this tool enhances property valuation accuracy in Georgia's dynamic housing market.

GBA C: Falling Money💸

This Game Boy Advance game, developed in C, challenges players to race against an opponent to collect falling cash and be the first to reach 25 points. The game features multiple screens including start, instruction, play, and end screens, with user input handling for movement and game control via arrow keys and button presses. Efficient sprite rendering, collision detection, and state management ensure smooth gameplay on GBA hardware.

Personal Portfolio💼

This portfolio website, built with Astro, serves as a creative outlet to showcase my projects and experiences while leveraging the benefits of React and static site generation. Hosted on Vercel with GitHub for version control, it reflects my evolving skills and design choices. Continuously maintained and updated, the site allows me to experiment with new ideas and express my creativity through web development.