Hongyu Jia (贾宏宇)

alt text 

Research Engineer
Baidu Online Network Technology (Beijing) Co., Ltd.
Beijing, China
E-mail: jiahy19@tsinghua.org.cn

A brief cv

About me

I received the M.S. degree from Tsinghua University, in 2022 and the B.S. degree from Dalian University of Technology, in 2019.
I am a core contributor to PaddlePaddle, Baidu's open source deep learning platform. Since joining Baidu, I have become the top contributor to PaddlePaddle on GitHub. My research focuses on AI for System, including topics like AI compilers, reinforcement learning, and deep learning frameworks. My goal is to advance deep learning by making it more accessible, efficient, and applicable to real-world problems. Through my work on PaddlePaddle, I strive to create innovations that democratize AI and unlock its benefits for all.

Education

M.S., Software Engineering, Tsinghua University, 06.2022

  • Awards: Glodon First-Class Scholarship 2020 (Top 10%), KDD Cup AutoML Track 4th place 2019

  • Main Courses: Deep Learning, Algorithm Analysis and Design, Mathematics for Engineering Master Candidate

B.S., Software Engineering (dual degree in Japanese), Dalian University of Technology, 06.2019

  • Awards: National Scholarship 2015 (Top 1%), School Outstanding Student 2016 (Top 1%)

  • Main Courses: Course Design of C/C++ Language, Probability and Statistics, Computer Organization and Structure

Publications

  1. Jia H, Jiang Y, Zhong C, et al. "TTDeep: Time-Triggered Scheduling for Real-Time Ethernet via Deep Reinforcement Learning." 2021 IEEE Global Communications Conference (GLOBECOM). IEEE, 2021. [pdf]

  2. Sun J, Jia H, Hu B, et al. "Speeding up very fast decision tree with low computational cost." Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence (IJCAI). 2021. [pdf]

  3. Zhong C, Jia H, Wan H, et al. "DRLS: A deep reinforcement learning based scheduler for time-triggered ethernet." 2021 International Conference on Computer Communications and Networks (ICCCN). IEEE, 2021. [pdf]

Work experience

  1. Senior Research Engineer (07.2022-Present), Baidu

    • The main contributor of the AI compiler within Paddle, a leading deep learning framework in China.

    • Developed equations between operators to ensure the correctness of operation fusion and loop fusion, aiming to surpass well-established compilers like TVM and Triton.

    • In charge of the custom operator mechanism, which helps millions of developers plug in Paddle more freely and quickly.

  2. Intern (07.2021-09.2021), Alibaba

    • Leveraged Bayesian networks to design an algorithm for constructing probabilistic causality diagrams, enabling analysis of key factors driving fluctuations in advertising sales metrics.

  3. Intern (03.2021-06.2021), Microsoft Research Asian

    • Designed subscribe-publish parallel training architecture; accelerated inverted index recall model by 32.5 times.

    • Optimized the space vector recall model and designed an asymmetric structure of the PolyEncoder algorithm.

  4. Intern (04.2020-08.2020), Tencent

    • Established relations between Wechat Reading’s users and books and adopted the GraphSAGE algorithm to optimize the DeepFM-based ranking model; increased the AUC index by 3%.