About Andrew
I am an entrepreneur and programmer, currently working as a software engineer at
CastTV.
My recent academic work has focused on the field of Intelligent Systems, specifically Artificial Intelligence in computer games and machine learning in web applications. I have also explored emergent phenomena in both nature and computer science.
I have 10 years of entrepreneurial development experience in Perl, MySQL, JavaScript, and HTML/CSS. I have built websites that leverage aggregated content, web scraping, and user input, and I have experience in Ruby, C/++, Python, and Java, as well as in unix administration.
Freebie Finder
Freebie Finder is an automated free stuff aggregator, receiving 3,000-4,000 visitors a day. Combining input from multiple sources, it automatically collects freebies from across the web while filtering out scams and referral pyramids. The software also attempts to predict freebies' categories and countries and then utilizes user input to refine these predictions. Freebie Finder is built upon open source software, including Perl, MySQL, Apache, and Lynx.
Research Experience
At Georgia Tech, I researched automated, personalized, adaptive tour guides in a museum setting using TTD-MDPs, a technique for generating a distribution of trajectories through a Markov Decision Process. We found that simulated tour guides reduced museum congestion while honoring museum visitors' preferences and autonomy.
For my Masters research project, I researched motion control of high degree-of-freedom manipulators, like octopus arms and elephant trunks, for animation and robotics. I explored local search and optimization in a real-world domain where random restarts, a standard technique, are unavailable. In Fall 2006, I developed a remotely hosted interface tool in JavaScript, Perl, and MySQL that analyzes both historical traffic flow through a website and a visitor's cursor motion in order to estimate a probability distribution over page links. High likelihood subsequent links are either enlarged or pre-selected, thus improving their Fitts' Law properties. During the summers of 2004 and 2005, I developed Java-based online physics pedagogy tools to aid in undergraduate education. I wrote full-featured numerical physics simulator to allow students real-time, hands-on experience with physical phenomena in an integrated environment and researched the effectiveness of these tools in a classroom setting. Find links to the applets here. During summer, 2003, I developed a low-cost, distributed data-processing cluster to look for radio pulsar signals. I independently developed and deployed parallel-processing software to coordinate signal analysis and distribute 600 GB of raw radio telescope data. I set up and administered the Linux computing cluster and wrote software to gather timing data for cluster efficiency analysis.
A. Cantino, F. Crawford, S. Dhital, J. Dougherty, & R. Sherman. A Low Cost Distributed Computing Approach to Pulsar Searches, Eleventh SIAM Conference on Parallel Processing for Scientific Computing, San Francisco, California, February 24-27, 2004.