Currently Teaching Assistant for Embedded System Development and Applications (CENG 4480)

We Are Hiring! We are always open for thesis/year/semester projects for self-motivated students, with the right mindset, to work with MASS@CUHK. If you are interested in Memory System Designs, In-Memory Computing Paradigms and Emerging Non-volatile Memory Technologies, get in touch!
Note: No Prior Requirements. We expect to customize research experiences for students and we consider it's our responsibility to learn and grow together with our mentees.

About Me

I am a first-year MPhil student in Department of Computer Science and Engineering, The Chinese University of Hong Kong. I am fortunate to be advised by Prof. Ming-Chang Yang as a member of Memory And Storage Systems Research Group (MASS). My current research interests lie on Memory Systems, Processing-In-Memory and Emerging Non-volatile Memory Systems, with a strong focus on DRAM and RaceTrack Memory.

I received my Honors BSc in Computer Science from The University of Nottingham Ningbo China, where I founded and served as the first Project Manager, in the User-Centric Computing Group. Along with my groupmates, we creat interesting insights, build novel prototypes and examine their real-world effects.

Current Research


  • Yaohua Wang, Lois Orosa, Xiangjun Peng, Yang Guo, Saugata Ghose, Minesh Patel, Jeremie Kim, Juan G√≥mez Luna, Mohammad Sadrosadati, Nika Mansouri Ghiasi, and Onur Mutlu,
    "FIGARO: Improving System Performance via Fine-Grained In-DRAM Data Relocation and Caching"
    Regular Paper in 53th ACM/IEEE International Symposium on Microarchitecture in October 2020, (MICRO'20).

  • Prior Research

    CHESTNUT: Fast, Scalable and Extremely Serendipitous Recommender Systems

    The concept of serendipity has been understood narrowly and we have previously extended it from only "unexpectedness + usefulness" to "relevance + unexpectedness + usefulness". CHESTNUT, from the prototype we built during 2017 to 2019, has provided the opportunity to examine the effectivenss and practicality of such an approach. Compared with other development toolkits (such as Apache Mahout), CHESTNUT is fast, scalable and extremely serendipitous among the largest data set publically. We announced CHESTNUT in [HCI'20] as an invited paper and cover its real-world effects from a large-scale user study, with 104 participants from 3 countries, in [HCI'20] as well.
    CHESTNUT Overview Image

    (HCI'20 CHESTNUT on Github)

    Rethinking Driving Simulations Infrastructure: Techniques and Extensions

    Me and my collegues have struggled with many aspects of current cognitive driving simulations, and make several attempts to address and mitigate these issues. We have constructed developer-specific Documentation and region-specific simulated suites (including both Scenes and Scenarios) in the past year; We further have attempted the synthesis between simulators and real-world video streams in [AutomotiveUI'19]; Based on the previous experiences, we have been bringing together the above lessons togther as Project O2. The key insight hereby is to provide effective supports for automated implementations. We separated them into Oneiros and Omniverse , to provide flexible, agile and easy-to-use integration for scene and scenario generations. Both works are currently under revisions and pending for further submissions.
    Conceptual Vid2vid View from OpenDS
    Tested Vid2vid View from OpenDS

    (AutomotiveUI'19 Video Synthesis)

    Conceptual Vid2vid View from OpenDS

    (CHI'20 OpenDS Characterization)

    BROOK: Multi-modal and Facial Video for Human-Vehicle Interaction Research

    Jointly supported by LandRover and Jaguar, me and my colleuges are working towards the BROOK database, as a domain-specific data support for Human-Vehicle Interactions. The key insight of BROOK is to provide consistent data streams from as many angles (sensors, cameras, coordinates and etc.) as possible. Every data stream could be viewed as the primary stream, and then use other streams to characterize and validate proposed designs/prototypes.

    Our goal is to provide essential building blocks for explainable designs of personalized, automated and adaptive Human-Vehicle Interactions. We have positioned BROOK in [CHI'20 Workshop] and announced our Face2Multi-modal prototype in [Automotive'20]. Our extensive characterizations and relevant designs would be announced later, and we would be happy to talk about them separately.

    BROOK Database Overview

    (BROOK Project Website)
    Current Mentees

    Mentoring contributes to my research in many ways, and I feel extrmely fortunate and excited to work with the following wonderful individuals, who have greatly helped me grow as either an individual or a researcher. It's tough to maintain the stimulating research enviroment but we are glad to make it work.

    Below it's an overview of our current and past research projects being active. We welcome any feedbacks and potential collaborations.

    Students@Chinese University of Hong Kong

  • Students@National University of Defense Technology
  • Zilin Song on Efficient Interpreter for General-Purpose DSP.
  • Shuolei Wang on Domain-Specific Languages for General-Purpose DSP.

  • Students@University of Nottingham
  • Wangkai Jin on

  • Services & Volunteers
      Conference Reviews
    • Reviewer in Asian-Pacific Design Automation Conference (ASP-DAC'2020).
    • External Reviewer in ACM Conference on Human Factors in Computing Systems (CHI'2020).
    • External Reviewer in ACM Conference on Automotive User Interfaces and Interactive
      Vehicular Applications (AutomotiveUI'2020, 2019).

    • Journal Reviews
    • Reviewer in ACM Journal on Emerging Technologies in Computing Systems (2020).
    • Reviewer in IEEE Transactions on Computers (2020).
    • External Reviewer in Springer The Journal of Supercomputing (2020, 2019).