Zhifei Yang (杨志飞)

I am a first-year master's student at the Department of Computer Science, ETH Zurich. I am broadly interested in computer systems research, especially in distributed systems, cloud computing, operating systems and programming languages. I am extremely excited about building underlying systems for emerging new applications, as well as designing and analyzing systems from a theoretical point of view.

I received my bachlor's degree in the Computer Science Experimental Class of Harbin Institute of Technology (HIT), Harbin, China in July 2017. Playing with all kinds of novel technologies while failing to go deep in any direction, I had been a member of the HIT IBM Technology Center (a.k.a. FoOTOo Lab) since the summer of my freshman year. Having studied as a semester exchange student in the Department of Computer Science, The University of Hong Kong (HKU) in my junior year, I found my great interests in systems and programming languages. During the final-year internship in the Systems Research Group of Microsoft Research Asia (MSRA), I researched on distributed machine learning systems with excitement under the guidance of Dr. Lintao Zhang, Dr. Jilong Xue and Dr. Hucheng Zhou, which made me determined to pursue an academic career and dream for a faculty position.

[CV] [中文简历] [GitHub] [LinkedIn] [Email]

04/16/2017 I have decided to go to ETH Zürich for further research and studies. Looking forward to the journey!
03/23/2017 We have achieved better LambdaMART training speed than LightGBM. We are preparing to open-source our GBDT system FastTree.
01/18/2017 We are working with the MS product group to transfer some optimization techniques on GBDT to the search ranking tool of Bing!
12/13/2016 I did some investigation on building wireless VR systems. Please find my research proposal here (1 page short version) or here (long version).

FastTree: Distributed Gradient Boosting Decision Tree Framework (Ongoing)

A distributed machine learning framework based on Gradient Boosting Decision Tree (GBDT, GBRT, GBM or MART). The current version is built on top of ChaNa, the RDMA-optimized distributed computing engine developed by MSRA Systems Group. With both system-level and algorithm-level optimizations, memory usage and communication cost are reduced, and the performance is better than state-of-the-art tools (e.g. LightGBM and XGBoost). A Spark version of the framework has been opensourced on GitHub as SparkTree.

Read More | SparkTree (Spark Version)

Distributed Gradient Boosting Decision Tree Learning Framework


Built atop the RDMA-optimized distributed computing engine ChaNa

Tree Ensembling Illustration

RISC-CPU with Multiple Interruption Support on Digilent Nexys 3 FPGA

Video More

Reads instructions from memory. Outputs to LCD screen. Written in VHDL.

RISC-CPU with Multiple Interruption on Digilent Nexys 3 FPGA

A C-like Language Compiler


Supporting all basic elements of C. Implemented a demo game with this language and OpenGL. Done with pleasure with Zoey.



Video More

Drive the rocket to catch the beats on time. Best Project in the course XNA Game Designing and Development. Done in honor with Nil.