I am an Assistant Professor and Y. T. Lo Faculty Fellow in Electrical and Computer Engineering, and an affiliated Assistant Professor in Computer Science at the University of Illinois at Urbana-Champaign (UIUC). I lead the Systems Platform Research Group. I received my Ph.D. in Computer Science from Georgia Tech in August 2017. During my Ph.D study, I specialized in building storage systems for platforms spanning from wearables to mobiles to large-scale data centers. My Ph.D. dissertation received the College of Computing Dissertation Award at Georgia Tech.

I have been working on memory and storage systems, with a focus on their access Interface, architecture Specialization, and development Approaches (the ISA of memory and storage systems). My research received the inaugural ACM SIGMICRO Early Career Award, NSF CAREER Award, NSF CRII Award, Dean's Award for Early Innovation, IEEE Micro Top Picks, USENIX Best Paper Award, and a few industry awards. Some of them have been transferred into products and featured by popular media agencies.

Most recently, I am superexcited to build sustainable AI infrastructures and platforms with computer architecture and systems techniques to accelerate the AI-driven knowledge discovery and retrieval. We also organized the first Workshop on Hot Topics in System Infrastructure (HotInfra'23).

I have 2-3 Postdoc openings in my research group on the following research topics: AI infrastructure, memory/storage architecture/systems, and data systems.

The ISA of Memory and Storage Systems

    Interface -- Enabling Byte-addressability Everywhere
  • G10 makes unified GPU memory/storage architectures truly usable for ML systems with smart tensor migrations (MICRO'23).
  • Our research DDP on the study of distributed data persistency models has been selected to IEEE Micro Top Picks'22.
  • DeepMC develops a framework that can validate NVM programs implemented with different persistency models (PPoPP'22).
  • DDP presents a comprehensive discussion on distributed data persistency models for distributed NVM systems (MICRO'21).
  • UniHeap proposes a new framework to manage persistent objects across different managed runtimes for NVM (SYSTOR'21).
  • Hoop presents an efficient out-of-place update approach for reducing the memory persistency overhead of NVM (ISCA'20).
  • P-INSPECT presents a hardware-assisted approach to checking persistent objects in NVM programs (MICRO'20).
  • AutoPersist creates a new programming model and an ease-of-use NVM framework for Java programming language (PLDI'19).
  • QuickCheck uses speculation to reduce the performance overhead of checking persistent objects in NVM frameworks (VEE'19).
  • FlatFlash presents a unified memory and storage architecture to exploit the byte-accessibility of solid-state drives (ASPLOS'19).
    Specialization -- Driving the Evolution of Modern Memory/Storage Systems
  • HARDES discusses how the SmartNIC device can accelerate the distributed transaction systems (ISCA'24).
  • RackBlox presents a software-defined rack-scale storage system with a network-storage co-design approach (SOSP'23).
  • Security War in File Systems shows our recent study on filesystem vulnerabilities (ACM TOS'23).
  • RSSD presents a hardware-isolated network/storage codesign approach to defend against new ransomware attacks (ASPLOS'22).
  • IceClave is the first trusted execution environment for securing in-storage computing and computational storage (MICRO'21).
  • FReaC Cache enables reconfigurable computing in the last level cache for accelerating near-data computing (MICRO'20).
  • DeepStore develops in-storage accelerators for supporting new and emerging deep learning based data queries (MICRO'19).
  • FS Vulnerability Study presents an empirical study of the most common security vulnerabilities in file systems (APSys'19).
  • Halo presents a near-cache accelerator solution to scale the flow classification and packet processing for NFV (ISCA'19).
  • SDC proposes a software-defined cache archtecture for efficient data indexing with a generic simple abstraction (ICS'19).
  • TimeSSD presents a new flash-based solid-state drive that provides the firmware-isolated time-traveling property (EuroSys'19).

Sustainable AI Infrastructure

  • TCloud investigates how the NPU architecture and its ISA should be upgraded for supporting NPU virtualization (MICRO'24).
  • Towards Building Sustainable AI Infrastructures with Modular Data Centers (NSF-WSCS'24).
  • G10 makes unified GPU memory/storage architectures truly usable for ML systems with smart tensor migrations (MICRO'23).
  • NeuCloud presents our recent study for enabling the system virtualization of NPUs in modern cloud infrastructures (HotOS'23).
  • V10 presents a hardware-assisted multi-tenancy framework for improving NPU resource utilization and fairness in cloud (ISCA'23).
  • Our research EVR on VR workload characterization and accelerator design has been selected to Micro Top Picks'20.
  • iSwitch proposes an In-Switch Computing system architecture to accelerate distributed reinforcement learning (ISCA'19).
  • EVR proposes an energy-efficient system architecture to accelerate the 360-degree video processing for VR (ISCA'19).

Full List of Publications