This May 2011, I am receiving a PhD in Computer Science at the University of Utah. I have been working as a graduate research assistant at Scientific Computing and Imaging (SCI) Institute for the last 6 years under the supervision of Professor Claudio Silva. I have worked on a number of different projects, including: feature detection, preservation, and reconstruction for point-based models; volume rendering of time-varying datasets; out-of-core isosurface extraction; high performance visualization, seismic data computation and exploration, PDE solvers on GPUs, and high-performance parallel computing for streaming architectures. My PhD thesis is about building a high performance parallel image processing framework for GPUs.

During my time at Utah, I also had the chance to do internships at major industrial labs. During Summer 2007, I was an intern in the Deep Computing group at the IBM T.J. Watson Research Center. That was my first experience in industry, and it was also the first time I worked with highly-parallel computing systems. My main task involved optimizing rendering algorithms on parallel architectures, in particular OpenGL-based algorithms, and a software ray tracer. Besides the graphics-related research aspects of the work, I was also introduced to sound software engineering practices, as the project was planned to be transitioned to be part of an IBM product.

This experience at IBM was directly applicable to my research when I returned to Utah. I became very interested in highly-parallel architectures, and I have been investigating new ways to exploit these architectures in GPUs through CUDA.

My second experience in industry was at the Quantitative Volume Interpretation Group at ExxonMobil's Upstream Research Company (URC) in Summer 2009. I worked on improving regularity computation and multi-attribute exploration for seismic data, and this work has succesfully been transitioned as product and it is now been used daily by scientists. The work at URC also had an effect on my dissertation work, in particular on ideas related to handling large volumes of data.

My dissertation addresses computational problems in coupled GPU/CPU systems. The framework that I developed works on GPU desktops, multi-GPU workstations, and GPU clusters. I had to build a complete image processing framework to exploit the processing power and bandwidth of this end-to-end system. Also, I worked on how to transform computations on irregular domains to regular domain (which are supported on GPUs); and how to perform out-of-core computation on GPUs efficiently. Our proposed computational technique on irregular domains is not only highly scalable on parallel machines but also robust and capable of dealing with a challenging image registration problem which involves very large deformation and unreliable tissue measurement.

I believe these hybrid CPU-GPU systems are the future of high-performance computing, and I plan to expand my work on scientific computing on these architectures. Also, I believe the target of Ph.D study is not to become an expert in your researching fields or to publish papers with great novelty and contribution, but about how to do research independently and cooperatively with people. The other aspects will arise naturally from this. For these reason, I would like to find a position that will give me a chance to work with people, to use my expertise to help people as well as to improve myself. It will be idea if the position will give me a chance to continue working on problems related to my expertise. Joining a top-notch team and facing with new challenge will always make me feel excited and energetic, and help me continue my training.