Research Map
HPC
Systems
Systems
GPU
Kernels
Kernels
Communication
Optimization
Optimization
AI
Systems
Systems
Networking
for AI
for AI
Distributed
Inference
Inference
To be
continued...
continued...
Performance x Scale
Featured Work
MLSys 2026 MAC-Attention
a Match-Amend-Complete Scheme for Fast and Accurate Attention Computation
IPDPS 2026 NIMBLE
From Skew to Symmetry: Node-Interconnect Multi-Path Balancing with Execution-time Planning for Modern GPU Clusters
MLSys 2025 DeepSpeed-FPDT
Training ultra long context language model with fully pipelined distributed transformer
IPDPS 2024 ExFlow
Inter-layer expert affinity for accelerating Mixture-of-Experts inference.
Research Areas
High-Performance Computing Networking for AI Distributed Inference GPU Kernels Communication Optimization Systems for ML
News
- May 28 I will be presenting my work “NIMBLE” at IPDPS 2026, New Orleans, LA.
- May 26 I will be joining Anyscale as research intern at San Francisco, working on Ray inference. Happy to grab a coffee!
- May 18 I will be presenting my work “MAC-Attention” at MLSys 2026, Bellevue, WA.
- Aug 14 I have concluded my internship at AI Frameworks team, Microsoft. Big thanks to Masahiro, Sam, and Walid for an extrordinary research experience!
Talks & Presentations
Open Source
Research Artifacts
Systems code, experiments, and reproducible project materials.
Interested in AI systems, GPU communication, or distributed inference?
yao.877@osu.edu