About

Hello! My name is Weixian Xu. I am a senior majoring in Computer Science at Shanghai Jiao Tong University, in the Zhiyuan College ACM Honors Class. I am currently a research intern at the SALT Lab at Stanford University, working with Prof. Diyi Yang. Previously, I was a remote research intern at the AI² Lab at Princeton University with Prof. Mengdi Wang, and a research intern at the GAIR Lab with Prof. Pengfei Liu.

I am always happy to talk about research and potential collaborations. Feel free to reach out.

Research Interests

I aim to develop increasingly intelligent systems to support practical applications and ultimately enhance human capabilities. A key strength of human intelligence lies in the ability to keep learning and growing. In contrast, as models face increasingly complex test-time environments that include intricate tasks and various human needs, their intelligence is still largely fixed after training.

My research therefore focuses on test-time learning, especially enabling AI to effectively identify what matters in increasingly heterogeneous feedback. I explore test-time learning across multiple directions.

Advancing AI itself

Draw insights from experiments to advance itself

Supporting practical applications

Meet diverse application requirements

Enhancing human capabilities

Infer users’ needs to better help humans

Real-world use Ultimate goal

Publications

Sherpa: Teaching LLMs to Teach Adaptively main figure

Sherpa: Teaching LLMs to Teach Adaptively

Weixian Xu, Yanzhe Zhang, Zora Zhiruo Wang, Changyu Chen, and Diyi Yang

Preprint

EEVEE: Towards Test-time Prompt Learning in the Real World for Self-Improving Agents main figure

EEVEE: Towards Test-time Prompt Learning in the Real World for Self-Improving Agents

Weixian Xu, Shilong Liu, and Mengdi Wang

2nd Workshop on Lifelong Agents (LLA), COLM 2026

ASI-Evolve: Towards AI-Driven Discovery Across the AI Research Stack main figure

ASI-Evolve: Towards AI-Driven Discovery Across the AI Research Stack

Weixian Xu‡, Tiantian Mi*, Yixiu Liu*, Yang Nan*, Zhimeng Zhou*, Lyumanshan Ye, Lin Zhang, Yu Qiao, and Pengfei Liu† (‡Leading author, *Core contributors, †Corresponding author)

Workshop on Language Models for Scientific Discovery (LM4Sci), COLM 2026

Neural Architecture Discovery via Autonomous Evolution main figure

Neural Architecture Discovery via Autonomous Evolution

Weixian Xu*, Yixiu Liu*, Yang Nan*, Lyumanshan Ye, Xiangkun Hu, Zhen Qin, and Pengfei Liu (*Equal contribution)

COLM 2026

Multi-Agent Evolve: LLM Self-Improve through Co-Evolution main figure

Multi-Agent Evolve: LLM Self-Improve through Co-Evolution

Yixing Chen, Yiding Wang, Siqi Zhu, Weixian Xu, Haofei Yu, Tao Feng, Muhan Zhang, Mostofa Patwary, and Jiaxuan You

EMNLP 2026

Speed by Simplicity: A Single-Stream Architecture for Fast Audio-Video Generative Foundation Model main figure

Speed by Simplicity: A Single-Stream Architecture for Fast Audio-Video Generative Foundation Model

GAIR & Sand.ai (contributing author: Weixian Xu; full author list in the paper)

Preprint

An AI System for Autonomous Algorithm Evolution in Drug Development main figure

An AI System for Autonomous Algorithm Evolution in Drug Development

Zhimeng Zhou, Yang Nan, Minjie Mou, Yuntao Qian, Yixiu Liu, Zhengyu Zuo, Hao Yang, Weixian Xu, Bo Li, Wanghao Jiang, Yanlin Ren, Yang Liao, Yimeng Wang, Yinghong Li, Qingxia Yang, Zhiheng Xi, Tiantian Mi, Huaicheng Sun, Pengfei Liu, and Feng Zhu

Preprint (bioRxiv)

Repositories

ASI-Evolve

871
Lead

A unified AI scientist framework that runs a learn–design–experiment–analyze cycle across data curation, model architecture design, and algorithm discovery.

ASI-Arch

1194
Lead

An AI scientist that autonomously designs linear attention architectures end to end, discovering 105 architectures that outperform Gated DeltaNet.

EEVEE

Lead

Test-time prompt learning that interleaves task routing and task solving to improve LLM agents on heterogeneous, real-world task streams.

Sherpa

Lead

Trains LLM teachers with multi-turn reinforcement learning to adapt their teaching to students with diverse learning preferences.

ACore

Personal Project

A comprehensive operating system kernel in Rust, supporting virtual memory and page allocation, process scheduling, thread management, user/kernel space isolation, and a virtual file system with multiple filesystems.

Imxc

Personal Project

A compiler for Mx*, a C-like language, written in Java. It reaches performance comparable to Clang through Mem2Reg and graph-coloring register allocation, with function inlining, SCCP, ADCE, and incremental multi-file builds.

RISC-V CPU

Personal Project

A RISC-V processor core in Verilog using Tomasulo’s algorithm, supporting the RV32I instruction set and deployment on FPGA hardware.

Background

Education

Shanghai Jiao Tong University
B.S. in Computer Science, Zhiyuan College (ACM Honors Class)
Sep 2023 – Jun 2027 (expected)

Research Experience

Stanford University · SALT Lab
Research Intern · Advisor: Prof. Diyi Yang
Jun 2026 – Present
Princeton University · AI² Lab
Remote Research Intern · Advisor: Prof. Mengdi Wang
Dec 2025 – Jun 2026
Shanghai Jiao Tong University · GAIR Lab
Research Intern · Advisor: Prof. Pengfei Liu
Jun 2025 – Jun 2026

Honors

Foresight-Sequoia Talent Development Fund
Top 5 students in Zhiyuan College
2025
Zhiyuan Honors Scholarship
Academic Excellence Award (top 2% in SJTU)
2024 – 2026
Provincial Second Prize, Chinese Physics Olympiad (CPhO)
Zhejiang Province
2022

Teaching & Service

Head Teaching Assistant
Data Structures (CS1951), Instructor: Prof. Yong Yu
Spring 2025
Teaching Assistant
Programming (CS1953), Instructor: Prof. Huiyu Weng
Fall 2024, 2025
Class Monitor, ACM Honors Class
Leading the ACM cohort, coordinating schedules and administrative tasks
Present