Portfolio

BigThink

Description
Big Think is a global online forum connecting people and ideas. Through an ever-expanding platform of knowledge content, including in-depth interviews with the world's leading experts, Big Think is a vital hub for important information to help you function, and succeed, in a rapidly changing world. In keeping with our belief that crucial information should be freely shared, discussed and debated, we have developed a full menu of tools to engage, disseminate, and subscribe to uniquely powerful content. Whether you use Twitter, Facebook, Digg.com, Delicious, Google Reader, Vimeo, YouTube, a personal blog, Tumblr, or any application with an RSS feed, Big Think allows you to share bright ideas with the wider Big Think audience as well as your personal cadre of lively thinkers—quickly and easily. (from BigThink)
Technlogies
Ruby on Rails, Javascript (Prototype), HTML, CSS, and SQL

Calipso Visualization in Google Earth

Description
The goal of this project was to create a tool capable of generating visualizations of CALIPSO data that were viewable on consumer PCs without the necessity of installing obscure software. CALIPSO generates profiles of clouds and aerosols in the atmosphere through an onboard LIDAR system. Google Earth provided compelling advantages for this sort of visualization, but it was clearly not designed to handle the Z-axis curtain files produced by CALIPSO's LIDAR instrument. We used the COLLADA 3D model format which is designed to represent buildings in Google Earth to erect 20km high walls onto which we projected the CALIPSO data. Image processing and navigational geometry were required to colorize the raw data, match the curvature of the Earth, and accurately curve the orbital path of each atmospheric slice. Our method was presented at the American Meteorological Society's 88th annual meeting. The current, optimized version of the tool was written in C++ using the Qt GUI toolkit. Due to the coarse multithreading and efficient data processing implementation, this utility can process a half gigabyte, half orbit HDF file in about 15 seconds on a quad core machine running at 2.4ghz.
Technlogies
C++, Ruby, Qt, Matlab, and KML

PixelNotes

Description
PixelNotes was my senior research project for high school. I created a note taking platform with Ruby on Rails that fed the class notes through different collective intelligence systems to try to identify important words that should be linked to other sources of information. The idea was to autogenerate inline links in the style of Wikipedia. The system recorded users' interactions with the autogenerated inline links to gauge their accuracy. The two algorithms I tested were Naive-Bayesian classifiers and Support Vector Machines. In addition to the human testing, I also wrote automated tests that compared the algorithms' results to the inline links in a random sample of Wikipedia articles. This test system was written in Erlang. I presented my results at the Virginia Junior Academy of Science 2008 Meeting and won first place in Computer Science for my work.
Technlogies
Ruby on Rails, C++, Qt, Erlang, Perl, SVMs, Bayesian Filters, JavaScript, HTML, and CSS