I'm a PhD student studying Artificial Intelligence at Northwestern University in the Qualitative Reasoning Group. My interests lie in the design and understanding of automated reasoning techniques that are applied to and can handle human-levels of knowledge. My most recent work involves the development of methods for encoding symbolic representations (e.g. first-order logic) as compact, fixed-length vectors (to be used as inputs to various statistical machine learning techniques). My thesis will describe how to design extremely data-efficient question-answering systems that can be applied to reasoning-intensive domains.
Sep 2015 - Present
- Pursuing PhD in Artificial Intellligence
- GPA - 4.0 / 4.0
- Advised by Ken Forbus
- Awarded Cognitive Science Fellowship
Aug 2011 - May 2015
- B.S. in Computer Science
- Minor in Mathematics
- GPA - 3.8 / 4.0
- Advised by David Leake
- Awarded the Dean's Advisory Council Senior Student Award
Jun - Sep 2018; May - Aug 2019
My primary focus while an intern at IBM was the development of neural-reasoning systems. In my first internship there, I wrote a superposition-based theorem prover that could be completely guided by deep-reinforcement learning techniques. Part of the work resulted in a patent application for discrete fixed-length vector-space encodings of structured data that were decodable. In the following summer, I developed method for encoding logical formulas with Graph LSTMs that resulted in significant performance improvements on premise-selection tasks. The LSTM encoding scheme also resulted in patent application.
May - Aug 2014
As an intern at Epic, I was tasked with writing a plagiarism detector that could automatically flag pairs of doctor notes with high degrees of similarity. The plagiarism detector was fairly simple, consisting of mostly trivial NLP similarity algorithms (e.g., Levenshtein distance).
May - Aug 2013
This was my first software engineering internship. I was on a team that was managing a data migration from Taleo to Peoplesoft. My objective was to develop a solution for streamlining grief report handling, thus helping to smooth the transition.