As a software developer in industry, I was frustrated by compilers which either do not expose or cannot really take advantage of novel hardware features, and written code in assembly language. I have also been bitten by bugs in complex multi-host control/data-collection/data-processing software systems which neither generations of testing and FDA approval, nor years of use in the field could wring out.As a university researcher, my goal is to fix these problems and show that the right abstractions can improve performance and productivity, resulting in provably correct numerical software, with the highest level of performance.As Chief Science Officer, my goal is to make sure these innovations are not just another academic curiosity, and that they are fully developed into commercially useful tools and processes.Specialties: High Performance Computation using SIMD/MulticoreSoftware Engineering (Provably Correct Optimization / Parallelization)Model-Based Magnetic Resonance Image ReconstructionInstruction Scheduling