I have been working on a number of projects over the years. I am gradually updating this page with my activity, so keep checking back. I'll also include screenshots and source code once I take the time to resurrect my sunsetted projects.
For my master's thesis, I developed a prototype for a system that employs a plug-and-play functionality to seamlessly integrate existing anomaly detection methods into healthcare audit databases. This system presents the analytical results to both administrators and patients in a way that facilitates the determination of the legitimacy of accesses to patient records.
I have developed a Python framework for statically generating the content of this website. See the github project for details.
With the help of two partners, we constructed a tool for detecting credit cards embedded within images. The motivation for this project was the observation that some people intentionally upload images of their credit cards to social networking sites. We presented and wrote a report on this project for a UIUC Machine Learning course.
I am currently the de facto database administrator for my research group. I have designed and implemented a normalized MySQL schema to hold our research dataset as well as a series of python scripts for importing and exporting data as well as using Enthought for statistical analysis.
I developed a JavaScript fuzzer designed to break JavaScript parsers/runtimes using a few thousand lines of C++. It is designed to produce semantically correct code so that the runtime can be fuzzed, rather than just the parser.
With the help of two classmates, I developed an application that extracts a set of accesses from an audit log (such as in an electronic health record) and constructs a workflow model from the unstructured data, à la Agrawal, et. al. Feel free to take a look at our presentation.
I built an EHR emulator which reads an EHR audit log, extracts workflow structures and emulates the behavior of the patients and users within the audit logs. The purpose of this was to have a rich set of data on which to run various classifiers (logistic regression as well as support vector machine with multiple kernels). I wrote this using a few thousand lines of C#. Check out my presentation here.
I wrote a compiler for the Classroom Object Oriented Language (COOL) for CS 426 - Compiler Construction. It was written in C and C++ and uses Flex for lexing, Bison for parsing, and LLVM as a backend. Everything else was either provided by the instructor as support code or written by myself. I also wrote an LLVM pass to perform register allocation for MIPS using the Chaitin-Briggs algorithm with the help of a partner.
I developed a graphing calculator for Windows using C#.NET. It supports multiple parsers should one wish to use a different syntax and is very extensible. It can graph functions f : R -> R, f : R -> R x R, and sets defined iteratively using complex values, like the Mandelbrot Set. It is also capable of symbolic differentiation, numerical integration, and several other features. I presented it to win 2nd place at the UTD programming competition.