Hello, I'm Kathy!

I'm a senior at the University of Pennsylvania poking around the tech space. Every now and then, I make some attempts to build cool things. As a student, my full-time job is to learn--often from textbooks, but more and more it is from talking, hacking, and/or eating with people much smarter than me.

I work at a computational genomics lab and TA a few CS classes (currently: a Haskell mini-course and a machine learning course) during the school year. My free moments are usually dedicated to revisiting good movies/TV shows (Friends!), tracking down great coffee and even better cake, and wandering around Philadelphia to admire the art (also to gape at the tiny dogs). When I am not woefully struggling to keep this site up-to-date, I am on the Internet for the cats.


Topic Modeling: A Deep Learning Approach
Python (theano, scikit-learn, nltk)
Leverages the Yelp Academic Dataset to answer the question: do topics commonly discussed in restaurant reviews differ by region? I was responsible for learning the specifics of the Theano deep learning classifier we wanted to use and then modifying the implementation to work with our input data and intended analysis. I also wrote a lot of the data pre-processing code (i.e. functions to filter data by location and restaurant type) and the post-processing extraction of topic "nouns" using nltk.
Designing a pathway-pathway interaction network
Python (pandas), R
The Greene Lab has recently developed and published a method called ADAGE that performs unsupervised integration of diverse biological experiments. ADAGE is completely data-driven, as opposed to knowledge-driven: the relationships that it discovers are not conditioned on prior knowledge. The goal of my project is to use this property of the model to identify curated pathways that frequently co-occur or exhibit mutual exclusivity. (For example, we'd expect oxidative phosphorylation and fermentation to exhibit exclusivity relationships, regulated by oxygen in opposite directions.)
JavaScript, MongoDB
This was developed by a 3-person team at MHacks IV Fall 2014. I was responsible for the client-facing aspects of the browser extension--that is, I implemented the functionality that allows a user to not only post their own comments, but also view and upvote/downvote existing comments. I also wrote the logic for serializing the data before it reached the database, as well as deserializing it when it needed to be rendered on the client-side.
Philadelphia Health IT Circle
Drupal, PHP, HTML, Sass/CSS
Commissioned by PHITCircle, a non-profit healthcare organization, this site was developed by a 4-person team during the final week of Philly Dev Camp. My task was to build a custom theme for this multi-page website. The layout is fully responsive (it is compatible on all screen sizes and devices) and formatted to suit the organization's branding.
Game: Walk the Line (2-D scrolling, platform)
Java, OOP
Objective is to traverse lines of an entire poem, avoid dying, and accumulate as many points as possible. Tiles generated with ASCII map showing where a poem should be "cut"; enemies can detect player's relative position after an attack; player has a fatigue mode triggered for a set time when ammo needs "reloading."


Software Engineering Intern, MongoDB performance team (2016)
hackNY Fellow (2016)
Technology Summer Analyst, Goldman Sachs (2015)
Penn IBI Undergraduate Research Fellow, Greene Lab (2015-present)
Download Resume

Contact Me

Send me a message!
You can also find me on github, facebook, linkedin, and twitter.