My Projects

Rally Ready!

Rally Ready! is a tool that I developed to help organize round robin tournaments. It allows you to upload an Excel file with player names and then generates a schedule for the tournament. The schedule is generated using a deterministic mathematical algorithm that ensures players have a fair number of matches and play with as many different opponents as possible.

SuperCalc

SuperCalc, or KenKen® or Kendoku®, are mathematical puzzle games that are similar to Sudoku®. The objective is to fill out a N by N Latin square while still adhering to the clues provided. Here is an example puzzle:
These puzzles are addictingly fun, and I decided to implement my own version of KenKen®, both as an exercise in development but also to add my own advanced features, such as more complex mathematical operations, the use of the numbers 0 to N-1 instead of 1 to N, and even hidden clues. If you are interested in playing an interactive SuperCalc puzzle, click here.

Weather or not!

Weather or not is a browser extension that three of my friends and I developed during the 2022 GT Hackathon. The idea is to integrate a weather service into Google Calendar, which can tell you the weather at a given events location.
This extension has not been published yet as it is in active development. The objective is to implement daily notifications that will inform you if any of your events will have inclement weather. However, if you are interested in installing the current version, check out our Github page for details and instructions.

Music Genre Classifier with Machine Learning

This was my teams final project for CS4641 Machine Learning in Fall 2023. We analyzed 1700 spectrograms of English songs between 270 and 300 seconds in length across 16 distinct genres. We utilized five different machine learning algorithms to examine our dataset: KMeans, SVM, Random Forest, Logistic Regression, and Convolutional NN. As we anticipated, our CNN performed the best, with an accuracy of over 70% on the test data. Additionally, we noted that all models performed well at distinguishing different types of classical music (excluding symphony and chamber), but most models experienced confusion between the R&B, Rock, and Indie genres, likely due to mitual inspiration from artists in those genres and also due to limitations in our dataset. Overall, we were happy with the performance of our models, and our final report can be seen here.