Tom Ostrand

Tom Ostrand is a Research Associate at Mälardalen University in Västerås, Sweden, and a Visiting Scholar at Rutgers University's Center for Discrete Math & Computer Science, where he is continuing his work in empirical software engineering.  Tom has contributed to the theory and practice of software testing, design of effective test strategies, and creation of innovative, productivity-enhancing test tools.  His main current interest is empirical studies of large software systems, with special emphasis on the prediction of faults in future versions of the software.  Tom has studied large software systems at AT&T, Siemens, and Univac.  He has analyzed the occurrence and types of software faults in those systems, and has created an interactive tool that predicts the parts of a system that are most likely to have faults in the next release of the system.  

Tom has gained experience at AT&T Labs, Siemens Corporate Research, Univac Software Research, and Rutgers University. During his career, he 

• Created the Category-Partition test design and specification method, one of the first formally defined approaches to systematic and thorough test specification. The approach was first implemented at Siemens Corporate Research, and a production version is now used by testing teams throughout Siemens Corp. 
• Participated in design and implementation of a tool to measure control-flow and data-flow based test coverage, and carried out an early experiment to compare effectiveness of different coverage metrics.  The subject programs created for that experiment became known as the Siemens suite, and have been a widely-used benchmark subject for evaluating software testing methods since 1995.
• Analyzed change and fault data, and constructed fault prediction models for many large AT&T software systems. Tom implemented the DEPICT fault prediction tool, a GUI-based tool that provides software developers accurate predictions of files most likely to contain bugs in future releases. 
• Was granted a patent for ATool for Predicting Fault-prone Software Files, U.S. Patent Number 8151146, April 3, 2012.
• Played a key role in elevating the Conference on Predictive Models in Software Engineering (PROMISE) from a local workshop to an internationally recognized professional conference, while serving as Program Chair and Steering Committee member in 2007, 08, 09.
• Was an active participant in the creation of the AT&T Software Test Engineer Professional Development Program.
• Did Sarbanes-Oxley compliance assessment for AT&T's sales and contracting processes, evaluated current procedures, designed and executed test scenarios, and made recommendations for procedure remediation.
• Taught graduate and undergraduate Computer Science courses at Rutgers University.