Low latency options trading.

Publications: * Morphological Priors for Probabilistic Neural Word Embeddings, P. Bhatia, R. Guthrie, J. Eisenstein EMNLP 2016.* Mimicking Word Embeddings using Subword RNNs, Y. Pinter, R. Guthrie, J. Eisenstein, EMNLP 2017.I worked under Dr. Jacob Eisenstein on deep learning approaches to lexical semantics. Our paper that appeared at EMNLP 2016 was a bayesian method for computing word embeddings, with sub-word information acting as the prior.Later work I did was continued work on learning high-quality representations for rare or out-of-vocabulary words for deep learning applications, especially relevant to low-resource languages, and some work on automatic hypernymy detection.

I TA'd an advanced undergraduate / graduate level course in Natural Language Processing. This involved creating a problem set where students implemented a state-of-the-art dependency parser (see projects below), and I also created a deep learning tutorial for this class that has been incorporated into Pytorch's official tutorials.I also TA'd for an introductory data structures and algorithms class in Java. I taught a 1-hour weekly recitation class, graded assignments, and held office hours. (Summer 2015).

I worked on the project Yeppp!, a low-level library for highly-optimized vector operations on the CPU. I wrote hand-optimized loops for high-throughput vector operations with modern SIMD extensions.

I worked on optimization of critical options pricing code for low-latency, where I achieved speedup as high as 20%, as well as several data analysis projects on volatility of European equities and futures that have options. Programmed primarily in C++ and R.

Developed an application for managing the assets of the IT department (e.g Windows keys, hardware) in ASP.NET MVC and C#.