I graduated from University of Delaware with a Ph.D. degree in Computer Science in Summer 2016.
My PhD advisor is Dr. John Cavazos.
My research area focused on energy-efficient high performance computing, machine learning methods to optimize HPC applications for performance and energy efficiency. I had extensive experience with energy management techniques involving DVFS and CPU Clock Modulation. I also applied Polyhedral Compilation and machine learning techniques to optimize and auto-tune applications for energy. Besides, my research involved GPGPU and sampling-based search techniques on large file systems.
I joined Intel Corporation in September 2016 after graduation.
We parallelize a cardiac arrhythmia simulation application using CUDA, OpenCL, OpenMP, and OpenACC. We achieved hundreds of times speedup over the sequential. Using OpenACC, we achieved almost the same speedup with minimal application code modification.
I modified a famous graph kernel (Weisfeiler-Lehman Subtree Kernel) and encoded a forest of directed trees into set of numbers and then derived the histogram of numbers for similarity calculation. Each number could be decoded back to the original tree structure.
We provide two larger applications than Polybench that can go through a polyhedral compiler: LULESH and Brdr2d -- the application used in the above project as well.
GLANCE does just-in-time analytics on large file systems. It was tested on a file system with more than 1 billion inodes. It applies random sampling technique and returns approximate number of files/directories in the file system without traversing the whole file system directory.
I have a Github page which hosts many other interesting projects.