Data Science Developer in Atlanta, GA, United States

A personal passion project developed and maintained a PyPi package complementing a computation linguistics paper published in Glottotheory (https://arxiv.org/pdf/1901.00521.pdf). The paper explores and refines ideas in information theory and NLP, particularly with regard to Heaps' Law.

I created this character-level N-gram language model package because I found the online tutorials for working with NLTK quite cumbersome, confusing, and slow. N-gram models are nowhere near the cutting edge of language modeling. However, they are still useful for establishing baselines, running simple information, like theoretic experiments and calculations, and overall useful for NLP research. This package is fast and truly plug-and-play.

I worked on this exciting math and optimization problem relevant to graphics, GPU computing, and geospatial analysis: What is the minimum aperture angle of a cone apex at the origin containing n random points in d-dimensional space? An even simpler version, do they all fall on one side of a (hyper)plane? This question represents a simple-sounding problem with very complex, inefficient dimension-specific solutions. Turns out to be a combinatoric, linear programming type problem, not necessarily geometric, and the established worst- and average-case solutions are not great.