Portrait of Mao Yang

Mao Yang

Principal Researcher and Research Manager


I (Mao Yang, 杨懋) received my Ph.D degree in computer science from Beijing University, China, in 2006. Before that, I received my B.S., M.S. in computer science from Harbin Institute of Technology, Harbin, in 2000 and 2002, respectively. Since 2006, I have been with Microsoft Research Asia, Beijing, as a Researcher in the System Research Group.

My research interests are in distributed systems, information retrieval systems, machine learning systems, and multimedia systems, especially for design, implement and deploy practical systems.

I am also an architect, and I worked on the following projects at Microsoft BING team:

  • The design and implementation of Cougar, a new ranking system for supporting the-state-of-art semantic ranking models. The system starts to serve all web queries since from 2013.
  • The design and implementation of Tiger, a new generation flash memory based index serving platform, and the system starts to serve all web queries since from 2012.
  • The design and implementation of replication and fail over protocol of Kirin, a new web store and processing system, and the system starts to process many billions of web data since from 2010.
  • Proposed a Web scale Q&A system that build into Web search engine. The system starts to provide directly answers in Bing since from 2016.

Some other research projects I’ve worked on include:

  • TLA Made Live: a formal method to build distributed systems.
  • The design and implementation of a large scale distributed storage system prototype PacificA. The protocol is also used by several open source projects, such as rDSN, Kafka.
  • Reconfiguration protocol for a paxos based replication state machine library.

My current research focus is on the AI infrastructure and Tools, and algorithms for Web search. We released several projects:

  • OpenPAI : An open source platform that provides complete AI model training and resource management capabilities, it is easy to extend and supports on-premise, cloud and hybrid environments in various scale.
  • NNI (neural network intelligence) : An open source AutoML toolkit for neural architecture search and hyper-parameter tuning.
  • MMdnn: A comprehensive, cross-framework solution to convert, visualize and diagnose deep neural network models.
  • Visual Studio Tools for AIVisual Studio Code Tools for AI and GitHub repo : An extension for Visual Studio and Visual Studio Code to build, test, and deploy Deep Learning / AI solutions.