Undergraduate at Sun Yat-sen University (CSE, 2023), member of ArcSysu Lab (Prof. Xianwei Zhang). Researching Agentic Systems and SysML. Passionate about turning ideas into reality and exploring all interesting projects.
Nexa is the first Harness Native Agent Language — where agent safety is enforced by the compiler, not by runtime frameworks. Write flows, not glue code. 16+ features, 1,500+ tests, Rust AVM runtime with WASM sandboxing.
Preprint | COLM 2026 Submission
Yipeng Ouyang, Xianwei Zhang
First Author
CRediT Roles: Conceptualization, Methodology, Project administration, Writing - original draft
Preprint | NeurIPS 2026 Submission
Yipeng Ouyang, Yi Xiao, Yuhao Gu, Xianwei Zhang
First Author
NeurIPS 2026
Yipeng Ouyang, Yi Xiao, Yuhao Gu, Xianwei Zhang
First Author
ACM CAIS 2026 Workshop AgentSkills
Yipeng Ouyang, Yi Xiao, Yuhao Gu, Xianwei Zhang
First Author
TACO
K. Wu, Y. Ouyang et al.
Co-author
Extension of GoPTX with additional tuning parameters and kernel matching strategies, achieving 15.8% overall performance improvement
CHI
K. Wu, H. Chen, Z. Zhu, Q. Lin, Y. Ouyang et al.
Co-author
AI/LLM module construction and key documentation writing
NeurIPS
Y. Han et al.
Non-author contributor
Layout and figure design
CVPR'25
Xinshuai Song et al.
Non-author contributor
Data screening and code review
Preprint
Zeming Wei et al.
Non-author contributor
Data screening and annotation
Undergraduate Researcher
Course Teaching Assistant
The first Harness Native Agent Language where agent safety is enforced by the compiler, not by runtime frameworks. Write flows, not glue code. 16+ features, 1,500+ tests, Rust AVM runtime.
The first compiler course integrating HPC + LLM source optimization. Top Grand Prize CCEC'25, First Prize CCEC'24, Best Poster Award DSI'24.
Extension of GoPTX with additional tuning parameters and kernel matching strategies, achieving 15.8% overall performance improvement. Submitting to TACO (CCF-A).
Exploring multi-agent argumentative generation and chain-of-thought as a novel compilation method, using domestic LLMs to drive C-to-Rust safe source code translation.
Using reinforcement learning for automatic compiler optimization on domestic supercomputer chips.
Mixed-precision compiler optimization targeting HUAWEI Ascend server (aarch64-openEuler).
Full-pipeline automated legal document identification and analysis using AI LLMs on supercomputing platform.
Rust-based Operating System for x86-64 architecture. Course project demonstrating OS fundamentals.
Exploration of thinking chain and interaction patterns with GPT-4o.
B.S. in Computer Science and Engineering
National Training Team Member
High School Diploma
Sole recipient in the CSE School
China Computer Education Conference, YatCC-AI team
China Computer Education Conference, YatCC-AI team
Workshop of Data Sharing and Infrastructure (China, Japan, Korea), December 2024. 1st place.
Novice category
Novice category
International Geography Olympiad, China regional
International Geography Olympiad, national training team
SYSU cybersecurity competition
Chinese calligraphy (running, cursive, regular script)
CSE School return-to-home practice photography contest