GregDefender

Volunteer Graduate Researcher

Tetherless World Constellation•  May 2018 - Present

Currently working as a volunteer on the Machine Common Sense project to create a semi-supervised language model that can extract new semantic triples from generic sentences in Allen Institute for AI’s GenericsKB database. During my internship in the summer of 2018, I was tasked with annotating MIMIC semantic data dictionaries to help Tetherless with their health informatics project and creating a research project of my own choosing. For my internship research project, I used NOAA weather data to train a supervised multiclass classification model and used k-nearest neighbor to predict weather forecasts. I used ARIMA modeling, Moving Averages, Polynomial and Linear Regression for my statistical analysis. After my internship, I worked on a research project with Professor James Hendler at RPI to use LSTMs in an AI system that formulates the responses in call and response songs.

Software Developer, Project Mentor

Rensselaer Center for Open Source•  January 2018 - January 2020

Worked on a music machine-learning open source project called Music ex Machina, whose purpose is to utilize Artificial Intelligence to create computer-generated music. Used Google's gensim and word2vec to determine a song's genre based on its title, singer, and album name. Experimented with LSTMs to create lyrics for jazz songs. Used Lillypond to create machine-readable jazz music and generate audio samples. Was an Open Source Mentor, and assisted new open-source members.

Software Developer Intern

The MITRE Corporation•  May 2019 - August 2019

Interned in MITRE's Advanced Technologies division for the Social Software and Collaboration department during the summer of 2019. Researched methods of implementing a Harvard Catalyst Profiles Server, an ontology-based application that helps researchers easily find collaborators and coauthored publishers, for my department’s genomics website. Worked on a MITRE internal research project called "Mining Media Reports for Information about Transportation Incidents." Over my internship, I established a proof of concept for using web mining to find train accident details and to send important findings back to our sponsors.