Personal Profile

Yang Luyan is an engineer at the Hetao Institute of Mathematics and Interdisciplinary Studies (Shenzhen), currently engaged in research and engineering work related to AI interpretability and AI Safety. She graduated with a bachelor's degree in Internet of Things Engineering from Beijing University of Posts and Telecommunications and a master's degree in Electronic Science and Technology from the Quantum Science and Engineering Research Institute of Southern University of Science and Technology.

The research and engineering directions focus on the analysis of internal mechanisms of large models, interpretability of neurons and representation layers, analysis of security alignment mechanisms, and engineering verification of security evaluation.

During my master's degree, I engaged in research on the application of deep learning in precision instruments and silicon-based quantum chip manufacturing. I have experience in abstracting complex scientific research problems and translating them into verifiable engineering systems. I have applied for invention patents for related achievements.

I once served as an AI engineer at Huawei Technologies Co., Ltd., where I participated in the construction of enterprise-level DevOps systems and large model-assisted development systems.Currently, I am primarily engaged in experimental design, system implementation, and engineering support in the fields of AI Safety and model interpretability.

 

Educational Background

2021.09 – 2024.06

Institute of Quantum Science and Engineering, Southern University of Science and Technology

Electronic Science and Technology

Master's degree


2016.09 – 2020.06

Beijing University of Posts and Telecommunications

Internet of Things Engineering

undergraduate

 

Work Experience

2024.08 – 2025.04

Huawei Technologies Co., Ltd

AI Engineer

 

Representative Achievements

Invention patent: A CNN-based method for classifying key feature images in the STM silicon-based direct writing process -- December 2023

Paper: Differentially Private Federated Learning via Alternating Direction Method of Multipliers -- 2023.12