Since 1999, I have been employed as a lead application software engineer in the transportation algorithms team at Manugistics Inc. in Wayne PA, working on transportation management software. In the algorithms team, we work on transportation optimization problems, all glorified and needlessly complicated versions of the Travelling Salesman Problem. Development environments run from old C/Unix code to Java/J2EE.
For three years before that (1996-1999), I was a senior applications software engineer in the Intelligence Systems group at Management and Data Systems, a Lockheed Martin Company, in Valley Forge PA.
Until 1996, I was a graduate student research assistant under the direction of Dr. Kang G. Shin in the Real-Time Computing Lab, at University of Michigan in Ann Arbor, MI. I defended my dissertation in April 1996, finished my revisions in July 1996, and received the long-awaited Ph.D. degree in December 1996. My dissertation investigated intelligent schedule execution and revision in a distributed manufacturing system.
My research interests include:
Scheduling
Among the buzzwords I can justify putting on my resume are: C,
C++, Unix, Java, J2EE/EJB, X/motif, Perl, SQL, Client/Server, Object-Oriented,
Structured Analysis, Artificial Intelligence, scheduling algorithm analysis
and development, satellite operations, satellite image simulation.
"Polite Rescheduling: Responding to Schedule Disruptions in a Multi-Agent
Manufacturing System."
The dissertation considers the problem of responding to unexpected events in a multi-agent (cellular) manufacturing system. Flexible manufacturing systems must be able to respond to unexpected events, including schedule disruptions. However, one manufacturing cell's response to a local disruption may adversely affect other cells in the system. In such a way, an initial disruption local to one cell may propagate throughout the system. In the approach we propose, an agent faced with a local disruption will try to respond in a way least disruptive to other agents, determining which response will be most appropriate in part through negotiation with other agents. This approach, which we cell "polite replanning", has the advantage of retaining much of any original distributed plan. We apply this concept to the domains of tool management and scheduling and job shop scheduling, and evaluate this approach through simulation. We also discuss PRIAM, a 'polite' rescheduling architecture which is currently under development, and discuss negotiation strategies and techniques relevant to this approach.