Zhengchun Liu (刘正春)

Senior Machine Learning Scientist at AWS AI Labs

Home Research Publication Experience Team News

Research Work Experience

Amazon Web Service (AWS) AI Labs, United States 2022.08 - Present

Machine Learning Scientist

Argonne National Laboratory, United States 2019.08 - 2022.08

Computer Scientist

  • Data Science and Learning for Computer System(Explain, Predict and Optimize).
  • Deep Learning in High Performance Computing environment.
  • The application of Artificial Intelligence for Science.

University of Chicago, United States 2018.03 - 2019.08

Research Scientist at the Computation Institute

  • Performance modeling and characterization of high performance computing system.
  • Design a scalable architecture for smart science ecosystems.
  • Explore methods for distributed and autonomous management of the systems.
  • Embed intelligence in relevant computer systems via machine learning

Argonne National Laboratory, United States 2016.09 - 2018.03

Postdoctoral Appointee, Advisor: Rajkumar Kettimuthu, Mentor: Sven Leyffer

  • Building robust analytic models for science at extreme scales
  • Modeling, simulating & optimizing for large data transfers over wide area networks– Simulating to explain the behavior of scientific workflows over a distributed infrastructure.
  • Designing cyberinfrastructure for on-demand scientific experiment data analysis.
  • Architecture, methods, and algorithms to support self-tune and self-manage science ecosystems.

Universitat Autonoma de Barcelona, Barcelona, Spain 2013.09 - 2016.08

Research Fellow, PhD candidate, Advisor: Emilio Luque

  • Modeling & simulating hospital emergency department using HPC and agent-based model.
  • Modeling & simulation for healthcare operations management.
  • Healthcare system operation data analysis and population aging study.
  • Model verification, model parameters calibration and model validation.
  • Optimization, Parallel programming, Agent-based modeling and simulation.

Oak Ridge National Laboratory, Tennessee, U.S.A. 2015.12 - 2016.04

Visiting Researcher, Host: Kalyan S. Perumalla

  • Performance modeling, verification and validation for scientific simulation on multi-GPU clusters.
  • Developed a framework template for efficient simulation on multi-GPU and multi-Core clusters.
  • Designed and developed a large-scale vehicle evacuation simulator on GPUs using CUDA.
  • Implemented an earthquake wave propagation model on multiple GPUs using CUDA.

Northwestern Polytechnical University, Xi’an, China 2006.09 - 2013.09

  • Designed and developed a rapid control prototype and testbed system for designing drones.
  • Design and implement flight control algorithms on real-time embedded control systems.
  • Developed a distributed, hardware-in-the-loop interactive simulation system for fixed-wing drone.
  • Learned and earned collaborative leadership.
  • Six-Degree-of-Freedom flight dynamics model of fixed-wing drone.

Engineering Work Experience

Software Engineer (Part-time) 2010.05 – 2013.06 | Xi’an FengLiTong Electronic CO. LTD

Application development on an ARM Cortex-M3 with μC/OS-II platform. More specifically, I was in charge of:

  1. μC/OS-II real-time operating system porting and public library development;

  2. USB driver and filesystem development for pen-drive drive in order to export historical records from the embedded device to an USB drive, and load configuration files and system firmware from the USB drive;

  3. OTA (On The Air technology) firmware upgrade, this subsystem handles the firmware upgrade request from the client side. It includes request and download firmware data from server and flash embedded device itself with new firmware (through In Application Program);

  4. Reliable communication via GSM network (GPRS);

  5. GPS module driver developing and location data parsing.

Backend software development of a communication server for millions of embedded devices, on Linux.

  1. This server works as a bridge for terminal devices and users, terminal devices report their real time information each minute (or even 15 seconds) and some unpredictable alarm events.

  2. The server needs to receive, store these information, and push subscribed events to user’s client.

  3. The users also can send some control and query requests to terminal devices through their clients, and the server needs to push these requests to the corresponding terminals and handle responses from the terminals. This service program runs on a distributed system.

Product (Outsourcing Service) 2010.09 – 2013.07

  1. Graphic HMIs (Human Machine Interfaces) based integrated power management for ambulances. This power management system consist of two parts, user-side HMIs and command executing embedded device. The main challeng includes high reliability, and PCB layout due to heavy current and high power. In this project, my work include embedded hardware design and firmware development, i.e., from customer’s requirements to the final production.

  2. I cooperated with the company directly to design a temperature control and management system for machine tool’s cooling oil. This embedded system was based on a 8051 MCU, system hardware includes sensors, user function keys, and display; software includes user operation response, sensor data processing and control logic.


Muchuan, Leshan, Sichuan, China

I grew up in this small, peaceful and beautiful village. It’s in the south-west part of China, and you can find its location on the Google map. My hometown is famous for Bamboo Sea. The forest coverage rate is as high as 77%, it is known as “natural oxygen bar”.


Crazy proverbs

Theory is when you know everything but nothing works. Practice is when everything works but no one knows why. In our lab, theory and practice are combined: nothing works and no one knows why.

                           
Prototype before polishing. Get it working before you optimize it.
HTML Counter unique visitors since March 2015