Xuanteng Huang's homepage | pronounciation: Seant (/sɪnt/), Chinese: 黃軒騰
myself
San Francisco, CA,
United States, Jun. 2025

I am a fourth-year Ph.D. student of arcSYSu at Sun Yat-sen University mentored by Assoc. Prof. Xianwei Zhang and Prof. Nong Xiao. My research interest mainly focuses on leveraging GPUs in machine learning systems.

I'm going to join NVIDIA RL team to develop frameworks for LLM post training, including but not limited to Megatron, TensorRT-LLM and NeMo RL.

I obtained my B.E. degree in 2022 from SYSU as well. Check out my résumé for more details.

Industry experiences

GPU compute arch intern
GPU compute arch internNVIDIA, RL team
Shanghai, ChinaMar. 2026 - Now
[object Object], ,recommandation system
Channel recommandation systemTencent, WeChat
Guangzhou, ChinaNov. 2024 - Jan. 2026
GPU compute arch intern
GPU compute arch internNVIDIA, F(ast)K(ernel) team
Shanghai, ChinaSept., 2022 - Dec., 2022
Heterogeneous intern
Heterogeneous internByteDance, IaaS virtualization team
Hangzhou, ChinaMay., 2022 - Aug., 2022

FOSS ? engagements

cpython
I've made some tiny contributions towards the main branch of CPython interpreter. I'm interested in the free-threaded no-GIL build and bytecode specialization optimizations to make Python faster. Also, I'm writing some articles about CPython internals in wiki/cpython.
debian
Since GSoC '24, I'm maintaining ROCm (open source GPU software stack for AMD GPUs) in Debian's official package archive. Unlike the package provided by AMD, we aim to provide a series of packages totally compatible with DFSG transparently for democratizing GPU computing. See my QA page for contribution records.