I am a software engineer at Diio, where I work on tools for sifting through sizeable aviation-related data sets.
Besides horseback riding and romantic walks on the beach, my interests include software testing, test adequacy criteria, software engineering education, and distributed systems. I also have an amateurish fondness for complexity theory.*
My... LinkedIn profile | Email is my first name followed by my last name, at yahoo.com
| Using Portfolio Theory for Better and More Consistent Quality | in ISSTA 2007 | |
| The effectiveness of software quality techniques varies. Many uncertain or unpredictable factors influence effectiveness, including human factors, the types of defects in the program, and luck. Compared to using a single quality technique, a diversified portfolio of techniques will typically be more effective and less variable. This work postulates a simple model, adapted from financial Modern Portfolio Theory, for the variability and effectiveness of techniques, singly and in portfolios. | ||
| State Coverage: A Structural Test Adequacy Criterion for Behavior Checking (joint work with David C. Kao) | in FSE 2007 | |
| We propose a new language-independent, structural test adequacy criterion called state coverage. State coverage measures whether unit-level tests check the outputs and side effects of a program. | ||
| A State Coverage Tool for JUnit | in ICSE 2008 | |
| We present a JUnit test runner that informs users of missing behavior checks in their tests. The tool tracks variable updates and definitions over the course of a test execution and determines which variables influence which assertions via dynamic taint analysis. | ||
| One of the projects I contributed to at Agitar Software was JUnitFactory.com, which automatically generated unit-level JUnit tests for Java code. There was no comprable tool for creating a safety net for existing (legacy) code without tests. Some generated tests are misses, but the hits are spooky good. |
| I previously worked on the open-source DARPA Bio-SPICE project as a member of Adam Arkin's lab at LBNL. Bio-SPICE integrates modeling and data analysis tools for cellular systems. |
| During the internet bubble, I worked for Echo Networks, a music startup which was acquired in 2003 by a consortium of music retailers. |
| My senior project (developed with Jim Brooks), the SmartSkipper, predicted the outcomes of baseball games based on a trained neural network. Its prediction rate on the 1999 season was better than Las Vegas bookies. (But not good enough to make money after the bookies' commisions.) It was trained on data from stats101.com (see below). |
| The summer after my junior year, I started stats101.com, an online sports statistics service, with two friends (James Deutsch and David Schwartz). It ran profitably for a year until we decided to get back to our textbooks. |
