Education

• Ph.D in King Abdullah University of Science and Technology, 2026 (expected)
• M.S. in The Chinese University of Hong Kong, 2022
• B.S. in Tianjin University, 2021

Work experience

• July 2025 - Present: Research Intern
  • Alibaba Qwen Team
  • Duties includes: Designed and implemented Canzona, a unified, asynchronous, and load-balanced framework to enable distributed matrix-based optimizers (e.g., Muon / Shampoo / SOAP) in large-scale LLM pretraining under Megatron with ZeRO-1 and Tensor Parallelism.
• Summer 2024: LLM Pretraining Engineer (Intern)
  • Aramco
  • Duties includes: Conducted large-scale LLM pretraining, optimized training efficiency through CUDA kernel fusion, asynchronous checkpointing, and distributed parallelism (ZeRO, DP, TP, PP).
• Fall 2022: Research Assistant
  • King Abdullah University of Science and Technology
  • Duties included: Distributed Federated Learning

Skills

• PyTorch / Libtorch: In-depth knowledge of PyTorch operators’ workflow and implementation, including distributed training packages, and multi-threading / streaming programming.
• CUDA programming / Triton: Intermediate in CUDA stream and kernel programming, with a solid understanding of CUDA principles.
• DeepSpeed / Megatron: Experience using DeepSpeed and Megatron for distributed training, including manual implementation for optimization.
• Programming Languages: Python (Mainly for PyTorch), C/C++ (Mainly for Multi-thread, CUDA Programming, and LibTorch).