I am an undergraduate student at Wesleyan University, pursuing a double major in Computer Science and Mathematics. As a researcher in Professor Thayer's bioinformatics lab at Wesleyan's College of Integrative Sciences, I specialize in developing analytical tools for molecular dynamics simulations, bridging computational techniques with real-world biological applications.
Developed a Python package for identifying inter-residue signaling patterns in proteins using probabilistic networks built from molecular dynamics simulations data. The project aims to provide deeper insights into how mutations and ligand binding affect signaling in proteins.
This poster, created for the Summer 2024 research symposium at Wesleyan, represents the starting point of my work in the lab. Since then, I've made significant progress and refined the project's direction. While the poster focuses on electrostatics, I am now incorporating additional interaction types. Currently, the project is in the testing and scientific validation phase. Future steps include integrating network analysis tools to pinpoint regions of the network responsible for shifts in signaling patterns. Identifying these regions could uncover potential targets for ligand binding, enabling restoration of native or near-native signaling. Visualization efforts are also underway, including heatmaps, graphs, and pathway representations using protein models from PDB files.
Implemented a web-based real-time tracking system for reusable food containers, including holder details, pickup/dropoff times and locations, etc. The system uses dynamic QR codes in dining halls to facilitate easy pickup and dropoff via an associated ios app. The project was started as a sustainability initiative for Wesleyan University dining system.
My three friends and I developed the entire system from scratch over an intense 24-hour period for a hackathon held at Wesleyan. We recorded this demo video literally 10 minutes before the projects were judged, which is why the quality isn't the best. That said, focusing all our effort on project development rather than cosmetics paid off: we won the internship prize track! Of course, an appealing look and the system's reliability are crucial. That is why we are currently working on converting this hackathon speedrun project into a robust, scalable, and maintainable product. We hope to deploy it by the end of Summer 2025.
Developed an album covers classifier using ResNet-50 (a pre-trained CNN) and transfer learning. The network's architecture was slightly adjusted, and multiple layers were unfrozen. It was then fine-tuned on a dataset of approximately 26,000 labeled album covers scraped from the internet. This is a new project still undergoing significant refinement. However, the latest test achieved a 74% accuracy in classifying rap, country, jazz, and classical albums.
I worked on this project as part of my audio-visual machine learning class. There's still more to do, including gathering additional data, applying class-specific data augmentation for genres that are more frequently misclassified, and experimenting with different models instead of ResNet-50. Once these steps are complete, I plan to deploy the project on the web.
I will be sharing some relevant updates about my work here every now and then.
Posted on December 8, 2024, 12:33 AM
Hey! I have finally finished my website!
Posted on December 8, 2024, 12:31 AM