The primary focus of Radu Grosu's research is to develop formal methods and tools which support the modeling and automated analysis of complex computational systems, including software systems, embedded systems and biological systems. The main emphasis is on approaches that scale well for realistic applications. My most notable contributions are in: Establishing a noncommutative Cayley-Hamilton theorem for finite automata; Showing that minimal nondeterministic finite automata may be related via linear transformations; Automatically detecting emergent properties in networks of cardiac myocytes; Automatically learning an efficient model for excitable cells; Defining a model checking technique that allows to trade time and space for precision and confidence; Defining compositional models for discrete and hybrid hierarchic automata, together with modular proof rules and search routines; Providing compositional semantics and refinement rules for UML sequence diagrams, and their automatic translation to statecharts; Providing an algebraic foundation of UML-RT in terms of trace categories; Giving a denotational semantics for dynamically reconfigurable systems.