Yuchen Xu

Ph.D. Student · Peking University

Email· Google Scholar dblp ORCID Semantic Scholar

I am a fifth-year Ph.D. student with the Institute of Data Science and Engineering, School of Computer Science, Peking University, under the supervision of Dr. Wenfei Wu. My research focuses on distributed systems for AI, including LLM training, inference, and multi-agent systems. I graduated from the School of Electronics Engineering and Computer Science, Peking University, with a Bachelor's degree in Computer Science and Technology, and also from the National School of Development, Peking University, with a double Bachelor's degree in Economics.

News

Sep 2026 AgentMesh accepted to ATC '26!
Aug 2026 Presented Turbo at SIGCOMM 2026.
Jun 2026 Homepage online.

Publications

(* Equal Contribution)

Multipath Collective Communication Beyond Scale-up Networks in GPU Clouds Yuchen Xu, Jianglong Nie, Baojia Li, Mingzhuo Chen, Hao Lu, Guanyu Qu, Zhenchuan Liu, Shuangshuang Yin, Xiaojie Huang, Chunzhi He, Yinben Xia, Quan Wen, Xiang Li, Zekun He, Yachen Wang, Xianneng Zou, Congcong Miao, Wenfei Wu.
21st European Conference on Computer Systems (EUROSYS '26), 2026
MimoSketch: A Framework to Mine Item Frequency on Multiple Nodes with Sketches Yuchen Xu, Wenfei Wu, Bohan Zhao, Tong Yang, Yikai Zhao.
Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD '23), 2023
Enabling Flexible and Efficient Collective Communication Scheduling in Distributed AI Yuchen Xu, Xiting Ju, Wenfei Wu.
2026 IEEE 32nd International Symposium on Local and Metropolitan Area Networks (LANMAN '26), 2026
Turbo: Efficiently Serving Long-Context Large Language Models with In-Network Aggregation Ying Wan*, Yuchen Xu*, Chuwen Zhang, Yingsheng Huang, Yong Feng, Wenquan Xu, Jialin Li, Mingwei Xu, Wenfei Wu, Congcong Miao.
ACM SIGCOMM 2026 Conference (SIGCOMM '26), 2026
MimoSketch: A Framework for Frequency-Based Mining Tasks on Multiple Nodes With Sketches Wenfei Wu*, Yuchen Xu*.
IEEE Transactions on Knowledge and Data Engineering (TKDE '25), 2025
Cuckoo Counter: Adaptive Structure of Counters for Accurate Frequency and Top-k Estimation Qilong Shi*, Yuchen Xu*, Jiuhua Qi*, Wenjun Li, Tong Yang, Yang Xu, Yi Wang.
IEEE/ACM Transactions on Networking (TON '23), 2023
EPIC: Abstraction and Polymorphism of In-Network Collectives on Ethernet Yitao Yuan, Jianglong Nie, Tianyu Bai, Ruizhe Zhou, Siyuan Cao, Xujie Fan, Yuchen Xu, Junkai Chen, Chenqi Zhao, Nengyuan Zhang, Shaoke Fang, Jiangyuan Chen, Yuanfeng Chen, Jiaqi Sun, Zhan Wang, Xiaohua Xu, Yuchao Zhang, Yang Liu, Xiangrui Yang, Jing Lin, Xiaohe Hu, Yang Li, Chao Jiang, Limin Xiao, Weifeng Zhang, Junjie Wang, Wei Cheng, Yazhu Lan, Jianbo Dong, Binzhang Fu, Wenfei Wu.
ACM SIGCOMM 2026 Conference (SIGCOMM '26), 2026
TurboTSS: A Packet Classifier with Fast Rule Lookup and Update for the Cloud Shaoke Fang*, Qingsong Liu*, Yuchen Xu, Weize Gao, Jianglong Nie, Wenfei Wu.
IEEE INFOCOM 2026 - IEEE Conference on Computer Communications (INFOCOM '26), 2026
INARouting: Efficient Multi-Job Routing Optimization for Hierarchical In-Network Aggregation Jianglong Nie, Yidan Yuan, Yuchen Xu, Yitao Yuan, Kehan Yao, Lu Lu, Xiaodong Duan, Wenfei Wu.
IEEE Transactions on Networking (TON '26), 2026
PRISM: A Workload-Predictive Job Scheduler for Clusters with In-Network Aggregation Capability Yidan Yuan, Jianglong Nie, Yuchen Xu, Yitao Yuan, Shaoke Fang, Wenfei Wu.
Proceedings of the ACM CoNext Workshop on In-Network Computing and AI for Distributed Systems (INCAS '25), 2025

Preprints

Accelerating Distributed Deep Learning using Lossless Homomorphic Compression Haoyu Li, Yuchen Xu, Jiayi Chen, Rohit Dwivedula, Wenfei Wu, Keqiang He, Aditya Akella, Daehyeok Kim.
arXiv '24, 2024
Fine-grained MoE Load Balancing with Linear Programming Chenqi Zhao, Wenfei Wu, Linhai Song, Yuchen Xu, Yitao Yuan.
arXiv '26, 2026

Research

LLM Training & Collective Communication

MPCCS (EUROSYS '26): Multipath collective communication with scale-up and scale-out networks, accelerating AI training.
FlexComm (LANMAN '26): Flexible inter-collective scheduling without deadlocks, accelerating AI training.

LLM Inference & In-Network Computing

Turbo (SIGCOMM '26): In-network scaling-reduce on programmable switches to aggregate attention blocks, accelerating long-context AI inference.

Agentic Systems

AgentMesh (ATC '26): Identify and remove redundancies and barriers in agentic dataflow, accelerating multi-agent systems.

In-Network Computing

EPIC (SIGCOMM '26): Ethernet Polymorphic In-network Collective, for AI training and inference.
INARouting (TON '26): Routing for hierarchical in-network aggregation, for AI training.
PRISM (INCAS '25): Predictive scheduler for in-network aggregation, for AI training.
LHC (arXiv '24): Lossless Homomorphic Compression algorithm with in-network aggregation, for AI training.

Data Stream Mining / Network Measurement

MimoSketch (KDD '23 and TKDE '25): An unbiased data structure and algorithm framework for distributed frequency-based data stream mining tasks, including item frequency estimation.
Cuckoo Counter (TON '23): A cuckoo-hash-based data structure for frequency and top-k estimation in network measurement.

Services

EUROSYS '26 Shadow Program Committee Member
IEEE Transactions on Dependable and Secure Computing (TDSC '25) Reviewer
Thirty-Fourth International Joint Conference on Artificial Intelligence (IJCAI '25) Emergency Reviewer