Publications
Peer-reviewed papers, ordered by year. Also on Google Scholar.
2026
- NeurIPS 2026SPIRIT: Speed-Driven Online Adaptation for Self-Speculative DecodingConference on Neural Information Processing SystemsTo appear
- USENIX ATC 2026Courier: Host-Transparent Direct Network I/O for Secure Containers in Cloud VMsUSENIX Annual Technical ConferenceTo appear
- SC 2026Quantifying the Cost of Software-Managed Synchronization on Multi-Host CXL MemoryThe International Conference for High Performance Computing, Networking, Storage, and AnalysisTo appear
- OSDI 2026vBOIDs: Taming Chaos via Coarse-Grained Scheduling Abstraction for Containers20th USENIX Symposium on Operating Systems Design and ImplementationAcceptance rate 19.9% (136/681) PDF
- OSDI 2026JANUS: Cross-World, Cooperative Nested Virtualization for Secure Containers20th USENIX Symposium on Operating Systems Design and ImplementationAcceptance rate 19.9% (136/681) PDF
- HPDC 2026Scaling Attention Beyond GPUs for LLM Inference35th International Symposium on High-Performance Parallel and Distributed ComputingAcceptance rate 12.5% (35/281) PDF
- EuroSys 2026AEP: Achieving Hierarchical Fault Tolerance in DSM Through Atomic Execution Protection21st European Conference on Computer SystemsAcceptance rate 18.5% (59/319, Fall) PDF
- ICDE 2026DLRover-LM: LLM Pre-Training Framework with Thousands of Accelerators in AntGroup42nd IEEE International Conference on Data Engineering
2025
- SYSTOR 2025Can Hardware Outsmart Software in Tiered Memory Management? A CMM-H Case StudyACM International Systems and Storage ConferenceAcceptance rate 37.2% (16/43) PDF
- USENIX ATC 2025LiteShield: Secure Containers via Lightweight, Composable Userspace μKernel ServicesUSENIX Annual Technical ConferenceAcceptance rate 15.8% (100/634) PDF
- USENIX ATC 2025DSA-2LM: A CPU-Free Tiered Memory Architecture with Intel DSAUSENIX Annual Technical ConferenceAcceptance rate 15.8% (100/634) PDF
- VLDB 2025mLoRA: Fine-Tuning LoRA Adapters via Highly-Efficient Pipeline Parallelism in Multiple GPUsProceedings of the VLDB Endowment
2024
- OSDI 2024Nomad: Non-Exclusive Memory Tiering via Transactional Page MigrationUSENIX Symposium on Operating Systems Design and ImplementationAcceptance rate 17.4% (49/282) PDF
- VLDB 2024DLRover-RM: Resource Optimization for Deep Recommendation Models Training in CloudProceedings of the VLDB EndowmentAcceptance rate 20% PDF
2023
- SOSP 2023PVM: Efficient Shadow Paging for Deploying Secure Containers in Cloud-native EnvironmentACM Symposium on Operating Systems PrinciplesAcceptance rate 18.7% (43/229) PDF
- HotStorage 2023When Caching Systems Meet Emerging Storage Devices: A Case StudyACM Workshop on Hot Topics in Storage and File SystemsAcceptance rate 39.5% (17/43) PDF
- USENIX ATC 2023P2Cache: Exploring Tiered Memory for In-Kernel File Systems CachingUSENIX Annual Technical ConferenceAcceptance rate 18.4% (65/353) PDF
- IPDPS 2023mFlow: Accelerating Packet Processing in Container Overlay Networks via Packet-level ParallelismIEEE International Parallel & Distributed Processing SymposiumAcceptance rate 25.7% (95/369) PDF
2022
- ICDCS 2022PRISM: Streamlined Packet Processing for Containers with Flow PrioritizationIEEE International Conference on Distributed Computing SystemsAcceptance rate 19.9% (114/573) PDF
2021
- PLOS @ SOSP 2021FlashCube: Fast Provisioning of Serverless Functions with Streamlined Container RuntimesWorkshop on Programming Languages and Operating Systems PDF
- EuroSys 2021Parallelizing Packet Processing in Container Overlay NetworksEuropean Conference on Computer SystemsAcceptance rate 19.9% (38/191) PDF
2020
- SoCC 2020BAOverlay: A Block-Accessible Overlay File System for Fast and Efficient Container StorageACM Symposium on Cloud ComputingAcceptance rate 24% (35/143) PDF
- CLOUD 2020FedMax: Enabling a Highly-Efficient Federated Learning FrameworkIEEE International Conference on Cloud ComputingAcceptance rate 20.7% (53/256)
- EuroSys 2020PLASMA: Programmable Elasticity for Stateful Cloud Computing ApplicationsEuropean Conference on Computer SystemsAcceptance rate 18.4% (43/234) PDF
- INFOCOM 2020SDN-based Order-aware Live Migration of Virtual MachinesIEEE International Conference on Computer CommunicationsAcceptance rate 19.8% (268/1354) PDF
2019
- HotCloud 2019Tackling Parallelization Challenges of Kernel Network Stack for Container Overlay NetworksUSENIX Workshop on Hot Topics in Cloud ComputingAcceptance rate 37.3% (22/59) PDF
- CLOUD 2019Exploring Potential for Non-Disruptive Vertical Auto Scaling and Resource Estimation in KubernetesIEEE International Conference on Cloud ComputingInvited
- IC2E 2019ShadeNF: Testing Online Network FunctionsIEEE International Conference on Cloud EngineeringAcceptance rate 26% PDF
- DCPerf @ INFOCOM 2019mWarp: Accelerating Intra-Host Live Container Migration via Memory WarpingIEEE INFOCOM WorkshopsInvited PDF
- VEE 2019Fast and Live Hypervisor ReplacementACM SIGPLAN/SIGOPS Conference on Virtual Execution Environments PDF
2018
- SoCC Poster 2018ShadeNF: A Platform for Online Network Function VerificationPoster, ACM Symposium on Cloud Computing PDF
2017
- Ph.D. Thesis 2017Rethinking Cloud Storage System Software under Multi-TenancyPurdue University
2016
- SoCC 2016BASS: Improving I/O Performance for Cloud Block Storage via Byte-Addressable Storage StackACM Symposium on Cloud ComputingAcceptance rate 25% PDF
- DSN 2016StorM: Enabling Tenant-Defined Cloud Storage Middle-Box ServicesIEEE/IFIP International Conference on Dependable Systems and NetworksAcceptance rate 22.3% PDF
2015
- SoCC 2015vFair: Latency-Aware Fair Storage Scheduling via Per-IO Cost-Based DifferentiationACM Symposium on Cloud ComputingAcceptance rate 21.6% PDF
- CLOUD 2015vHaul: Towards Optimal Scheduling of Live Multi-VM Migration for Multi-tier ApplicationsIEEE International Conference on Cloud ComputingAcceptance rate 14% PDF
2013
- Middleware 2013Hybnet: Network Manager for a Hybrid Network InfrastructureACM/IFIP/USENIX International Middleware Conference (Industry Track)Acceptance rate 22.7% PDF
- USENIX ATC 2013vTurbo: Accelerating Virtual Machine I/O Processing Using Designated Turbo-Sliced CoreUSENIX Annual Technical ConferenceAcceptance rate 14.2% PDF
2012
- VEE 2012Virtualization Challenges: A View from Server Consolidation PerspectiveACM SIGPLAN/SIGOPS Conference on Virtual Execution EnvironmentsAcceptance rate 37.7% PDF
2010
- IISWC 2010Tackling the Challenges of Server Consolidation on Multi-core SystemsIEEE International Symposium on Workload CharacterizationAcceptance rate 37.5%
- EURASIP JASP 2010Subgraphs Matching-based Side Information Generation for Distributed Multiview Video CodingEURASIP Journal on Advances in Signal Processing
2009
- ICC 2009Graph Matching Based Side Information Generation for Distributed Multi-view Video CodingIEEE International Conference on Communications
2008
- ISCAS 2008Side Information Generation with Constrained Relaxation for Distributed Multi-view Video CodingIEEE International Symposium on Circuits and Systems
2007
- ICME 2007A Source-Driven Error Recovery Scheme using Wyner-Ziv CodingIEEE International Conference on Multimedia and Expo