Scholar
Shiyi Cao
Google Scholar ID: NR59_v4AAAAJ
UC Berkeley
Citations & Impact
Citations
1,106
H-index
11
i10-index
12
Publications
19
Co-authors
8
list available
Publications
Resume (English only)
Academic Achievements
- MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs (ASPLOS 2025)
- GraphPipe: Improving Performance and Scalability of DNN Training with Graph Pipeline Parallelism (ASPLOS 2025)
- Fairness in Serving Large Language Models (OSDI 2024)
- S-LoRA: Serving Thousands of Concurrent LoRA Adapters (MLSys 2024)
- Accelerating Data Serialization/Deserialization Protocols with In-Network Compute (ExaMPI@SC 2022)
- AdaM: An Adaptive Fine-Grained Scheme for Distributed Metadata Management (ICPP 2019)
Background
- Mainly interested in accelerating/optimizing computations (especially ML workloads) on large-scale heterogeneous systems. Currently looking for research interns interested in efficient RL training systems or agent training.
Miscellany
- Personal interests include efficient RL training systems and agent training.
