Second International Workshop on Software Engineering for Computational Science and Engineering May 23, 2009 Abstract: In order to make significant progress in solving challenging problems in Computational Science & Engineering (CS&E), we need to integrate a large amount of software written by different groups of experts. Modern Lean/Agile methodologies would seem to provide a good foundation for research-driven development of complex CS&E software. Here, we describe issues related to the integration of CS&E software and propose different integration processes tailored to the special challenges in CS&E. We also describe practical experience with some of these tailored integration strategies related to Trilinos and some of its important application customers at Sandia National Labs. Roscoe A. Bartlett Department of Optimization & Uncertainty Estimation Trilinos Software Engineering Technologies and Integration Lead Sandia National Laboratories