George Almasi is a Research Staff Member at IBM T.J. Watson Research. He has split his time more or less evenly between HPC and cloud computing. Looking for the mythical easy-to-program-but-high-performance HPC programming environment, he evolved from MPI and messaging libraries through array languages to UPC, co-array Fortran, X10 and the Asynchronous Partitioned Global Adress space programming model. The goal is still out of reach.
A first approximation for a cloud clusters is that they are HPC systems with major scaling, security and systems management problems and a chaotic usage pattern. George is spending most of his time on scaling and systems management.
George holds a PhD in Computer Science from the University of Illinois at Urbana Champaign (2001).