A young HPC + AI infra enthusiast

Haoquan Chen
陈淏泉

I build the infrastructure layer between ambitious workloads and the heterogeneous machines that run them.

Research Interests

HPC SystemsSystem software, scheduling, and runtime support for large-scale scientific computing.
GPU ArchitectureMulti-GPU execution, accelerator memory systems, and architecture-aware performance.
Distributed AI InfraTraining, serving, scheduling, and resource management for AI clusters.
Performance PortabilityCompiler and container layers that adapt applications across heterogeneous platforms.

Contact

  • Emailchenhq79@mail2.sysu.edu.cn
  • GitHubgithub.com/Nickchen-PUSH
  • Websitechen-hq.site

Education

Sun Yat-sen University

B.S. in Computer Science and Technology, School of Computer Science and Engineering.

Competition World Champion ISC25 Student Cluster Competition
Research SC'25 · CCF-A coMtainer, compilation-assisted HPC containers
Systems work Real machines Lineshine, Tianhe, Sunway, Bridges-2 ...

Experience

How I learned systems from real supercomputers

Labs taught me abstractions; supercomputers taught me constraints; competitions taught me to measure.

MiniMax

Developing a Megatron-based training framework for frontier video models, scaling PyTorch workloads across large GPU clusters while improving parallel efficiency, communication performance, and fault recovery.

MegatronPyTorchNCCLDistributed Training

arcSYSu Lab, Sun Yat-sen University

Worked around HPC containers, compiler-assisted deployment, and distributed system support for heterogeneous clusters. Built practical experience with container runtimes, compiler toolchains, reproducible evaluation, and production systems on Tianhe-Xingyi.

HPC ContainersDockerKubernetesGCC/LLVM

SYSU Supercomputing Team

Full-stack HPC training from the ground up: hardware setup, cluster configuration, operator development, GPU tuning, application optimization, and performance profiling, with hands-on experience on self-built clusters and Bridges-2.

Parallel ProgrammingGPUProfilingBridges-2

AIR, Tsinghua University

Worked around large-scale AI for Science applications and the system infrastructure behind real parallel workloads, connecting AI frameworks, MPI-style execution, scientific simulation, and deployment constraints across Tianhe-3, Sunway, and Tecorigin.

AI4SPyTorchMPISimulation

Technical Stack

ProgrammingPython, C/C++, CUDA
Parallel ComputingMPI, OpenMP, SIMD
PerformanceNsight, VTune, uProf, Perf
HPC & CloudDocker, Kubernetes, Slurm
AI InfraMegatron, PyTorch, NCCL

Research, Projects, Awards

What came out of the experiments

Artifacts from turning systems ideas into papers, prototypes, and evaluations that survive real machines.

SC'25 · CCF-A

coMtainer: Compilation-assisted HPC Container Images with Enhanced Adaptability

Co-developed an HPC container framework and automated validation pipeline for heterogeneous clusters, preserving 97% native performance while delivering up to 253% speedup over traditional deployment workflows.

coMtainer framework
Performance Optimization

OpenMX

Refactored compute kernels, introduced hybrid MPI + OpenMP parallelism, and profiled on Bridges-2 to achieve a 2.2x end-to-end speedup.

MPIOpenMPBridges-2
GPU Acceleration

OpenFOAM

Optimized CUDA C++ kernels and NCCL communication with Nsight-guided profiling, delivering a 1.5x overall speedup.

CUDA C++NCCLNsight
HPC Platform

YatCC-OL

Built a Kubernetes-based HPC application platform that served more than 200 students on the Tianhe-Xingyi supercomputer.

KubernetesRedisDocker
Recognition

Awards

  • World ChampionISC25 Student Cluster Competition, June 2025.
  • Top PrizeTencent Scholarship, ranked 1 / 424, Nov 2025.
  • Third PrizeCompiler Design Competition 2024.
  • Outstanding StudentsSun Yat-sen University scholarship, 2023-2025.