Builder, Woodworker, and Photographer
Shot on a DJI Mavic Mini.
Hi, I’m Phillip! I’m a junior studying CS and Math at Columbia University. I most recently interned as a ML engineer at Expedition Technology.
I’m interested in computer vision, building cool products, and math modeling.
This summer, I interned as a ML engineering intern for radio frequency analysis at Expedition Technology. I worked on computer vision based signals classification, ultimately building a hybrid Mamba architecture based model.
This past spring, I worked as a SWE at Axpo, building out tools for energy trading including automated data pipelines for fetching and preparing time series data and a bot to report and automatically annotate interday portfolio trade changes.
Freshman summer, I interned at Jangle Inc. software development team. There, I worked on a project integrating Jangle's product with Zotero’s 7.5 million users, allowing them to access, edit, and upload changes on Jangle’s reading application with duplication and version control.
Designed a print function to produce a physical copy of digital guidebooks.
In my freshman spring of 2023, I worked on a team of two other freshman to build a spatial-temporal differential equations model on the changing species composition of a plant biome through cycles of drought. Using a custom made algorithmically randomized rainfall model alongside randomized initial seeding, we ran simulations of a sample biome space with millions of data points over several years. The paper was awarded the MAA (Mathematical Association of America) award and outstanding Winner (top 6 teams for the chosen problem) out of 10,000+ teams at the International MCM 2021. Paper accessed here.
To practive for the MCM, our team also built an optimization model to identify NYC zip codes which would benefit the most from bike land investment based on maximal usage and community benefits. Paper accessed here.
As a musician, I've found sight-reading to be a crucial skill to develop, but the opportunities to find truly new music are inherently limited. Aimed at early-intermediate level musicians, I built the backend of Polyhymnia, an application which uses AI and Markov Chains to algorithmically generate unlimited original sheet music. The application also transfers the Needleman-Wunsch algorithm for standardized music alignment to judge student recordings, adjusting the difficulty in future generated sheet music based on performance. Github Repo
Through the Inspirit AI summer program, I compared traditional neural network architecture with that of CNNs and implemented transfer learning using VGG16, ultimately devloping an image recognition algorithm using TensorFlow to identify cars and people in videos from car windshields.
An educational simulation on income taxes which allows users to compare tax policies in several countries, estimate income taxes given a user’s salary, and gain a better understanding of tax policy formation through thought-provoking situational studies.
I am also an avid photographer and videographer in my free time and have previously worked gigs for in restaurants, events, and real estate. Here's a sample drone video a friend and I made of Princeton's campus.
My passion for creating doesn’t just stop at software projects but also translates to real world, where I have spent the past summer learning and practicing my woodworking skills. All of my projects are made by hand with nothing more than a whittling knife. Here are some of my favorite pieces.
I believe that the cognition and reflection inherent in the constructive and original nature of writing crowns it as the ultimate form of expressing ideas. I write occasionally on Medium on topics in technology, policy, and sociology. Some of my most viewed articles include “The Genius of Plant-Based Meat”, “Unconscious Bias: The Power and Risks of Societal Narratives”, and “Nuclear: Our Last Best Option”.
Questions, comments or thoughts? You can reach me at pmy2105@columbia.edu.