I am a PhD Candidate at University of California, Riverside. Prior to the PhD study in Computer Science, I obtained MS and BS degrees from Columbia University and Peking University.

News

• Mar 2025: Gave a talk at the Las Vegas, NV (PPoPP 2025).
• June. 2023: I gave a talk at International Conference on Supercomputing 2023.
• April. 2023: A paper was accepted at International Conference on Supercomputing 2023.

Education

• Ph.D. in Computer Science (Sep. 2022 – Present)
  University of California, Riverside (UCR)
  Advisor: Prof. Zizhong Chen

• M.S. in Electrical Engineering (Sep. 2020 – May 2022)
  Columbia University

• B.S. in Computer Science (Sep. 2016 – Jul. 2020)
  B.S. in Economics (Double Major)
  Peking University

Research Experience

1. USC ISI / Argonne National Laboratory (Jan. 2024 – Present)
   Los Angeles, CA / Lemont, IL
   Scientific Workflow Applications on Resilient Metasystem
   Mentors: Dr. Franck Cappello, Dr. Sheng Di, Dr. Krishnan Raghavan (ANL); Dr. Ewa Deelman (USC ISI)
   • Designed a Q-learning + GNN-based topology protocol (DGRO) that reduces network diameter by optimizing virtual rings over heterogeneous, failure-prone systems.
   • Implemented a single-hop gossip-based failure detector, resilient to network jitter and churn, enabling decentralized membership monitoring across 20+ globally distributed sites.
   • Deployed DGRO on the FABRIC testbed spanning Japan, Europe, Hawaii, and 15+ U.S. locations, demonstrating fast convergence and robustness at international scale.
2. UCR / Lawrence Berkeley National Laboratory (Sep. 2022 – Present)
   Riverside, CA
   Data-driven Exascale Control of Optically Driven Excitations in Chemical and Material Systems
   Mentor: Dr. Zizhong Chen
   • Designed and implemented in-kernel ABFT GEMM using tensor cores, achieving higher performance than cuBLAS while ensuring fault detection and correction under soft errors.
   • Developed a fully GPU-resident ABFT FFT pipeline, outperforming cuFFT, and enabling error-resilient spectral analysis in scientific simulations.
   • Proposed the first ABFT-enabled K-means clustering framework on GPUs, exceeding cuML performance with integrated resilience support.
   • Innovated lightweight, low-overhead in-kernel fault tolerance mechanisms across linear algebra and ML workloads, demonstrating resilience-performance co-design in exascale systems.
3. Nvidia (Jun. 2024 – Sep. 2024)
   Santa Clara, CA
   Compiler Optimization for OpenMP Target Offload on Heterogeneous GPU Architectures
   Mentor: Dr. David Appelhans
   • Investigated performance bottlenecks of OpenMP target offload in SPEChpc 2021 on GH200/H200 GPUs.
   • Developed compiler/runtime optimizations achieving up to 10× speedup without source code changes.
   • Analyzed OpenMP vs. OpenACC performance and contributed optimized versions to SPEChpc 1.1.9.
   • Work adopted by RWTH Aachen University, demonstrating both research impact and practical relevance.
4. Columbia University / AI4Finance Foundation (Aug. 2021 – Jul. 2022)
   New York, NY
   ElegantRL: Massively Parallel Deep Reinforcement Learning Library
   Mentors: Dr. Xiaoyang Liu, Dr. Xiaodong Wang
   • Developed multi-agent RL algorithms in ElegantRL, a popular RL library with ~4k GitHub stars.
   • Co-led ElegantRL_Solver, a high-performance solver that outperforms Gurobi for dense MaxCut problems.

Selected Publications

Full list in Google Scholar