Liwei Guo
Logo Assistant Professor, Doctoral Advisor

I am a tenure-track Assistant Professor at University of Electronic Science and Technology of China (UESTC) in the school of Information and Software Engineering. I received my Ph.D. from the University of Virginia, advised by Prof. Felix Xiaozhu Lin. My research is in the intersection of OS x Arch x PL, under the umbrella of efficiency and security.

Prior to UESTC and the facultyship, I was lucky to be awarded Top Minds from Huawei, where I had a little dabble in industry.

What I do:

  • Old-fashioned, old-school OS kernels
    • an in-depth analysis of Rust-for-Linux (USENIX ATC'24, Best Paper Award (first in China))
    • a built-from-scratch software virtual machine with an inverse guest/host paradigm (USENIX ATC'19)
    • an OS abstraction which provides energy isolation (EUROSYS'18)
  • Trustworthy computing
    • device drivers which run fully inside Arm TrustZone (EUROSYS'22)
  • Resource-efficient LLM
    • an elastic IO/compute pipeline for efficient transformer inference (ASPLOS'23)

Recently exploring:

  • System support for AI security
    • LLM, computer vision
    • Embodied AI, agents
  • New kernels for emerging scenarios
    • Confidential computing
    • Cloud gaming, rendering
    • Fully heterogeneous SoCs

I am always looking for motivated students who are interested in above topics (esp. in the hardcore part). If you'd like to join our group (as an undergrad intern or a graduate student), feel free to shoot me an email with your cv!


Education
  • University of Virginia
    University of Virginia
    Ph.D. in Computer Science
    2020 - 2022
  • Purdue University
    Purdue University
    M.S. in Computer Engineering (direct Ph.D. program)
    2016 - 2020
  • Sun Yat-sen University
    Sun Yat-sen University
    B.S. in Software Engineering
    2012 - 2016
Honors & Awards
  • USENIX ATC Best Paper Award (The FIRST in China)
    2024
  • Huawei TopMinds
    2023
News
2025
Site up! Happy New Year
Feb 05
Selected Publications (i.e. those I would like to be on my tombstone) (view all )
An Empirical Study of Rust-for-Linux: The Success, Dissatisfaction, and Compromise
An Empirical Study of Rust-for-Linux: The Success, Dissatisfaction, and Compromise

Hongyu Li*, Liwei Guo*, Yexuan Yang, Shangguang Wang, Mengwei Xu (* equal contribution)

USENIX Annual Technical Conference (USENIX ATC) 2024 Best Paper Award

We shed light on how Rust as an emerging memory-safe (systems) programming language blends with Linux kernel -- an old, giant legacy.

An Empirical Study of Rust-for-Linux: The Success, Dissatisfaction, and Compromise

Hongyu Li*, Liwei Guo*, Yexuan Yang, Shangguang Wang, Mengwei Xu (* equal contribution)

USENIX Annual Technical Conference (USENIX ATC) 2024 Best Paper Award

We shed light on how Rust as an emerging memory-safe (systems) programming language blends with Linux kernel -- an old, giant legacy.

Transkernel: Bridging Monolithic Kernels to Peripheral Cores
Transkernel: Bridging Monolithic Kernels to Peripheral Cores

Liwei Guo, Shuang Zhai, Yi Qiao, Felix Xiaozhu Lin

USENIX Annual Technical Conference (USENIX ATC) 2019

We present a novel OS model for heterogeneous multi-processors, which executes unmodified energy-consuming kernel code of a CPU on a microcontroller-like peripheral core. By combining DBT and emulation, a transkernel bridges ISA gaps and caters to core asymmetry with an inverse guest/host virtualization paradigm. Transkernel is inspired by various video game emulators.

Transkernel: Bridging Monolithic Kernels to Peripheral Cores

Liwei Guo, Shuang Zhai, Yi Qiao, Felix Xiaozhu Lin

USENIX Annual Technical Conference (USENIX ATC) 2019

We present a novel OS model for heterogeneous multi-processors, which executes unmodified energy-consuming kernel code of a CPU on a microcontroller-like peripheral core. By combining DBT and emulation, a transkernel bridges ISA gaps and caters to core asymmetry with an inverse guest/host virtualization paradigm. Transkernel is inspired by various video game emulators.

STI: Turbocharge NLP Inference at the Edge via Elastic Pipelining
STI: Turbocharge NLP Inference at the Edge via Elastic Pipelining

Liwei Guo, Wonkyo Choe, Felix Xiaozhu Lin

International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) 2023

We show how large Transformer models can run on a wimpy, cheap SoC at 1MB of memory while maintaining accuracy.

STI: Turbocharge NLP Inference at the Edge via Elastic Pipelining

Liwei Guo, Wonkyo Choe, Felix Xiaozhu Lin

International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) 2023

We show how large Transformer models can run on a wimpy, cheap SoC at 1MB of memory while maintaining accuracy.

All publications