PORTFOLIO
Below are selected projects that showcase my work across machine learning, software engineering, data analysis, and technical design.
01
CheXpert Competition
Collaborated on a machine learning team project focused on automated anomaly detection in chest radiographs using the CheXpert dataset. Developed predictive models to identify key clinical findings such as Cardiomegaly, Lung Opacity, and Pleural Effusion across thousands of chest X-rays. Employed advanced ML techniques to minimize mean squared error (MSE) across multiple diagnosis labels, with scores normalized by category variance. Gained hands-on experience with medical imaging data, evaluation pipelines, and collaborative model development in a high-stakes, real-world healthcare context.
02
Unsupervised Machine Learning for Analysis of Young Star Photometric Light Curves (2024)
I analyzed photometric light curves of young stars using multiple unsupervised machine learning techniques to further understanding of early star formation processes under the guidance of Professor Lynne Hillenbrand in the PMA department. In this project, I Worked with data from the Kepler/K2 and Transiting Exoplanet Survey Satellite (TESS) NASA missions, which include high-precision photometry for thousands of stars in the K2 dataset and over 200,000 stars in the TESS dataset to extract relevant features.​
02
An Investigation of the Robustness of Machine Perception for Control
Modern day control systems such as autonomous driving typically utilize image and video signals to produce feedback control. However, adversarial noise can often interfere with label prediction and thus lead to decreased robustness. This project aims to investigate adversarial robustness of convolutional nerual network (CNN) based machine perception for identifying the position of a driving car.
03
Tower Hopper (Platformer Game)
Designed and developed a fully functional 2D platformer game in a team of 3 where a treasure hunter ascends a dynamic tower by jumping between walls while avoiding enemies and obstacles. Implemented game physics including gravity, collision detection, and surface friction types (normal, icy, sticky) to affect movement. Built enemy AI (ghosts) with chase behavior and damage mechanics with temporary invincibility. Used sprites and polygonal level design, with keyboard-based controls for gameplay navigation.
03
Fire Detection Using Keras (2022)
For my AP Research project in my senior year of high school, I successfully designed and trained a convolutional neural network able to identify forest fires from an image using Keras in Google Colab. I then further implemented this model so that it could be used to process videos.
