Work/Projects

My Github handle is ozziefallick.

Papers

I've been working with some roommates on Choron, which is a chore market for shared households. Choron allocates chores to residents who dislike each chore least, and creates a currency that residents earn by doing chores. This currency can be transferred between residents in exchange for favors.

From the summer of 2011 to the summer of 2013, I interned at Arcode Corporation, originally through UMD's Corporate Scholar Program. While at Arcode, I worked on Inky, a smart email client.

From the fall semester of 2011 to the spring of 2014, I worked on a machine transliteration project for an honors thesis under Hal Daumé III. In this project, I used carmel, a finite-state automaton evaluator, trained on Wiktionary and Wikipedia data, to automatically transcribe words from script A to script B, while replacing sounds from language A that are unpronouncable in language B with their best approximations in language B. As a result, if A and B use the same script, then it returns language B's best approximation of the word.

For my Digital Cultures and Creativity capstone project, I made a Java program to convert written oratory to a MIDI interpretation. I included a GUI to allow the user to create their own mapping between International Phonetic Alphabet transcription and musical notation.

During my freshman year, I worked on a program in Java to visualize speech cadence in written language under Tanya Clement.

During the Software Design course in my senior year of high school, I was team leader on a 3-D space racing game called orbit. In orbit, the player controls a comet that loses mass when far from planets and gains mass when nearby. The player must pass through a series of waypoints to complete each race, using the gravity of nearby planets to pick up speed. The game was written in Java using OpenGL for the graphics.

As a Senior Research Project in high school, I researched and constructed a model for ancient Western warfare, corresponding approximately to the period between 500 BCE and 500 CE. The user would set up a battle using the GUI I made, and the battle would be simulated step-by-step to determine a winner. The model achieved approximately 90% success at predicting the winner (without any evaluation for magnitude of victory) on the battles I tested it with.