OSDI '26 Poster Session

Communication as a First-Class Scheduling Constraint in Distributed LLM Training
Vrishab Nooka, UNC Charlotte; Liz Morrell, The Urban Institute

Automatic Optimization of Distributed Protocols
David Chu McElroy, UC Berkeley

WasmBounds: Eliminating Memory Bounds Checks in WebAssembly
Emma Sudo and Keith Winstein, Stanford University

Prepack: Efficient Offline Microbatch Packing for LLM Pretraining
David McDermott, University of Michigan

Runtime Memory Management for On-Path SmartNICs
Kshitij Rana, UT Austin; Landon Johnson, University of Illinois Urbana-Champaign; Daehyeok Kim, UT Austin

Enabling Multitenant Kernel Extensions with ChameleonOS
Ting He, Julie Wang, Timothée Zerbib, Franco Solleza, Milda Zizyte, and Malte Schwarzkopf, Brown University

Toward Practical Remapping-based TCP Zero-copy Receive
Kisu Kim and Sihoon Seong, Sungkyunkwan University; KyoungSoo Park, Seoul National University; Jaehyun Hwang, Sungkyunkwan University

Rearchitecting RDMA-based Remote Storage Stack by Decoupling Target-side I/O Processing
Taegon Park, Yunjae Jo, Chanseo Park, Taeyoung Park, Beomseok Nam, and Jaehyun Hwang, Sungkyunkwan University

Model Intervention as a Service
Zikai Wang, Jaden Fiotto-Kaufman, Adam Belfki, Michael Ripa, Gabriele Sarti, Emma Bortz, Cheng Tan, and David Bau, Northeastern University

Verifying a distributed register implementation with Verus
Baltasar Dinis, Upamanyu Sharma, Nickolai Zeldovich, and Frans Kaashoek, MIT

Owl: Handling Cross-Service Performance Issue in Production Cloud Systems
Wenbo Qian, Boston University; Yile Gu, University of Washington; Yuhan Yao, Microsoft; Baris Kasikci, University of Washington; Ze Li and Murali Chintalapati, Microsoft; Yigong Hu, Boston University

Beyond Bounds Checks: Reducing Overheads in Embedded WebAssembly
Nicole Mulvey, Stanford University; Leon Schuermann, Princeton University; Zachary Yedidia and Philip Levis, Stanford University

AutoMan-Verus: Auto-Generating Verified Distributed Systems at Native Performance
Zihao Zhang, Ze Tang, and Shuai Mu, Stony Brook University

Jetpack: Consensus Made Generally Fast
Ze Tang, Zihao Zhang, and Weihai Shen, Stony Brook University; Jicheng Shi, DatenLord Inc.; Shuai Mu, Stony Brook University

Ahead-of-time Analysis of Shell Program Effects
Lukas Lazarek, Evangelos Lamprou, Zekai Li, and Nikos Vasilakis, Brown University

Efficient & Adaptable Detection of Malicious LLM Prompts via Bootstrap Aggregation
Shayan Ali Hassan, KAUST; Zaeem Mohtashim Khan, LUMS; Tao Ni, KAUST; Zafar Ayyub Qazi, KAUST and LUMS; Marco Canini, KAUST

Make Congestion Control Easy with Trafik
Alexander Khosrowshahi, Edward Nguyen, Franco Solleza, and Akshay Narayan, Brown University

Rest Easy: Constraining Untrusted Agent Tool Calls with Haven
Justus Adam, Yuchen Lu, Deepti Raghavan, Malte Schwarzkopf, and Nikos Vasilakis, Brown University

Co-Location as a First-Class Strategy for Microservice Deployment
Sebastian (Teng) Jiang, Robert Stanley, and Woyu Wang, University of California, Los Angeles; Ratul Mahajan and Arvind Krishnamurthy, University of Washington; Konstantinos Kallas and Sam Kumar, University of California, Los Angeles

A Kernel Native Approach to High Performance Networking
Ayesha Naeem, Eric Munson, and Orran Krieger, Boston University

Microservices! But at what COST?
Oğuzhan Çölkesen, Chloe Qiao, Alex Wang, Akshay Narayan, and Nikos Vasilakis, Brown University

MetaServe: Decoupling Policy from Mechanism in Heterogeneous AI Workloads
Dhruv Garg, Ada Gavrilovska, and Alexey Tumanov, Georgia Institute of Technology

Staggered Training for Compound AI Reinforcement Learning
Morgan Borjigin-Wang, Nayani Modugula, Ethan Lavi, Po Hao Chen, Akshay Narayan, Deepti Raghavan, and Malte Schwarzkopf, Brown University

HEDGE: Resource-Efficient and Lossless Expert Reduction for MoE Inference
Vima Gupta, Georgia Tech