Dr. Sergei Shudler

- Investigate task-based in-situ analysis and visualization techniques in extreme-scale scientific applications - Implemented analysis modules in C++ to provide users a richer set of features. - Explore performance analysis techniques for multicore processors and GPUs.

- Worked on SENSEI, a generic in-situ data analysis platform based on C++, and published a paper in EGPGV'19 focusing on one of its use cases. - Implemented analysis routines in Python and Numpy for a summer datacamp. - Experimented with OpenCL and OpenMP 4.5 offloading API in an effort to port user code to a new HPC system.

- Designed a framework that helps users validate the scalability of their HPC code. Published two peer-reviewed papers describing the results: ICS'15 and TPDS '19. - Implemented a lightweight benchmark for MPI in C++ and used it to profile performance. - Proposed an experimental method to understand the relation between parallelism, input sizes, and efficiency. Published results in PPoPP'17. - Developed a tool in C++ that allows users to understand contention and parallelism in OpenMP applications.

- Worked on a C++ and OpenGL-based 3D visualization system. - Introduced a multithreaded, progressive fetching mechanism for multi-resolution visual data to improve UI responsiveness. Collaborated with other developers to integrate it into the system. - Developed a functionality to display semi-transparent, floating text annotations within an OpenGL 3D scene. Used Win32 to port this functionality to Windows.

- Implemented (using OpenCL) an image denoising algorithm based on the Haar wavelet transform. Source code available in my Github (DeNoising). - Ported CUDA code to OpenCL as part of a preparation for a 3-day course.

Parallel programming and High-Performance Computing