Xinyin Ma
PhD Research Intern at NVIDIA
Singapore, Singapore
Summary
🤩
Rockstar🎓
Top SchoolXinyin Ma is a PhD candidate in Electrical & Computer Engineering at the National University of Singapore and a PhD Research Intern at NVIDIA with nine years of industry experience bridging research and applied ML. She has interned at Alibaba DAMO Academy and NetEase, contributing to production-focused research in computer vision and model optimization. An active open-source contributor, she helped extend the widely used torch-pruning library—adding multi-input/output support and transformer pruning (LayerNorm, Embeddings) and fixing tricky channel-rounding bugs. Her background combines strong academic credentials from Zhejiang University with hands-on engineering that moves cutting-edge pruning techniques from prototype to robust tooling. Colleagues describe her work as detail-oriented and impact-driven, often surfacing subtle implementation issues that materially improve library reliability.
9 years of coding experience
Master of Engineering - MEng, Computer Science, 3.93/4, Master of Engineering - MEng, Computer Science, 3.93/4 at Zhejiang UniversityDoctor of Philosophy - PhD, Electrical & Computer Engineering, Doctor of Philosophy - PhD, Electrical & Computer Engineering at National University of Singapore
Github Skills (4)
pytorch10
model-compression10
transformers9
python9
Programming languages (3)
Github contributions (5)
VainF/Torch-Pruning
Mar 2021 - Sep 2021
[CVPR 2023] DepGraph: Towards Any Structural Pruning
Role in this project:
ML Engineer
ML EngineerContributions:6 commits, 1 PR in 6 months
Contributions summary:Xinyin primarily contributed to the development and maintenance of the `torch-pruning` library. Their work included implementing support for models with multiple inputs and outputs, as well as integrating pruning functionalities for transformers, including LayerNorm and Embedding layers. They also fixed bugs related to channel rounding, ensuring the library's robustness.
pytorchnetwork-pruningpruningaccelerationmachine-learning
horseee/LLM-Pruner
May 2023 - Aug 2024
[NeurIPS 2023] LLM-Pruner: On the Structural Pruning of Large Language Models. Support Llama-3/3.1, Llama-2, LLaMA, BLOOM, Vicuna, Baichuan, TinyLlama, etc.
Contributions:3 PRs, 74 pushes, 2 branches in 1 year 2 months
baichuanbloomchatglmcompressionlanguage-model
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.