Ph.D. in Computer Science at Rutgers University


Hello! Thanks for your interest and visiting my website.

I'm now a Ph.D. student in Computer Science department of Rutgers University, and my advisors are Zheng Zhang and Ulrich Kremer. My current concentration is Operating System, Computer Architecture, Compiler & Programming Language, Parallel Computing and Cyber-Physical Systems. I'm also the co-founder of a startup on Cyber-Physical System application of cycling Bikerules Inc. From 2013 to 2015, I was working as a full-time software engineer in Fujitsu developping firmware of ETERNUS Disk storage system for 2 years after my bachelor's degree in Software Engineering at Nanjing University of Posts and Telecommunications (NUPT).

I'm quite familiar with low level stuffs in computer science such as computer micro-architecture, operating system, compiler, complexity. And I'm also eager to learn Approximation and Machine Learning stuffs. I'll periodically update my bloggers and post new articles.

If you like my bloggers, have amazing research project or some great open source projects that I could participate, feel free to contact me. Here is my resume. And if you're also interested in participating in developing or investing on Bikerules, do not hesitate sending an email to me :)

My email address is:

My Books Collection

One of my hobby is to collect books that I like. The following are part of the books that I collected in 2 years.


"DP2GV" Workstation

I recently built a hexacore CPU with dual GPU powerful monster for parallel computing research and deep learning stuffs. Xeon E5-1650 v4, in my opition, strikes the perfect balance between frequency and core numbers. GTX 980 is not as good as GTX 1080 but GTX 980 is very inexpensive on ebay. For multiple GPU system build, I strongly recommend Nvidia reference card. Supermicro mobo do not support SLI and I don't even need SLI for I don't play video games. 32GB ddr4 ecc reg memory is pretty sufficient for current use. Intel 750 pcie ssd has incredible I/O speed which hold the whole system while all the training data are put onto hard disk.