Distributed systems, grid and collaborative computing with emphasis on data intensive distributed computing, resource allocation and management, fault tolerance, coordination of computation and I/O, distributed data scheduling and management, use of distributed systems to solve problems in biology, astronomy, high energy physics, and coastal and environmental sciences.
He is the primary designer and developer of the Stork distributed data scheduling system which has been adopted by many national and international institutions. Dr. Kosar has spent three summers at IBM T.J. Watson Research Center doing research on several state-of-the-art topics, such as on-demand computing, storage area networks, and web capacity planning.
He has published two book chapters and many journal and conference papers. Dr. Kosar has served on the technical program committee of several conferences and workshops, on NSF panel reviews, and on many journal and conference peer reviews. He gave invited talks at several national and international forums.