Building reliable scientific-computing infrastructure and exploring the next generation of weather intelligence.
氣象專業 × 高效能運算 × 人工智慧
I am a meteorologist and HPC system architect at Taiwan's Central Weather Administration, working at the intersection of operational meteorology, numerical weather prediction, and large-scale scientific computing.
My background is in atmospheric science and cloud microphysics, with experience using MM5 and WRF. I previously led global data-assimilation work for more than a decade, contributing to ensemble typhoon-track prediction and hybrid methods including 3DEnVar and 4DEnVar.
Today, I focus on reliable, scalable HPC and NWP systems while exploring how machine learning, deep learning, LLMs, and retrieval-augmented generation can strengthen weather forecasting, disaster response, and scientific operations.
Weather-trained. Infrastructure-minded. AI-forward.
- HPC operations — Slurm, GPU systems, storage performance, monitoring, and production reliability
- Numerical weather prediction — data assimilation, ensemble prediction, WRF, MPAS, and UFS
- AI for weather — domain corpora, RAG, LLMs, and GPU-accelerated training
- Scientific software — Fortran, C/C++, Python, CUDA, MPI, OpenMP, OpenACC, and Spack
- Knowledge sharing — practical HPC education for scientific-computing teams
| Project | What it does |
|---|---|
| SovereginAI | Builds a structured Traditional Chinese corpus and training pipeline for Taiwan-focused weather AI. |
| CWA-HPC-tutorial | Beginner-friendly HPC workshop covering Fortran, C++, A64FX/SVE, CUDA, OpenACC, MPI, and practical optimization. |
| slrum-toolbox | Slurm operations and analytics toolkit for cluster health checks, usage reports, job workflows, and troubleshooting. |
| storage-bench | Reproducible FIO-based storage benchmarks with Slurm submission and CSV result aggregation. |
| CWA-MPAS | MPAS-JEDI build notes, portability fixes, debugging workflows, and MPI performance investigations. |
| ChurchAttend | Dockerized Flask and MongoDB attendance system with responsive UI, reporting, and operational tooling. |
I am interested in practical collaborations where atmospheric science, AI, and high-performance computing meet.
讓氣象科學、運算基礎設施與 AI 真正接軌。