Research
Our main research directions. Each page gives an overview of the topic, the people working on it, and our publications in the area.
Data Center ML InferenceMaking ML inference systems scalable, efficient, and sustainable in data centers through request routing, resource scheduling, autoscaling, and caching.
- FA2 (RTAS 2022)
- LambdaGrapher (WWW 2024)
- InferCool (SoCC 2024)
- Janus (IPDPS 2025)
- vLLM Benchmarking (MLSys 2026)
In-Network ComputingAccelerating application components on network devices including programmable switches and SmartNICs through programming frameworks, management systems, and algorithms.
- HIRE (ASPLOS 2021)
- C3 (HotNets 2021)
- NetCL (SC 2024)
- FluxReduce (SoCC 2026)
Mobile Agentic SystemsGetting sophisticated agentic workflows to run on mobile/edge devices, with privacy and efficiency, through benchmarking, kernel optimizations, and resource scheduling.
- Jellyfish (RTSS 2022)
- Uirapuru (MobiCom 2025)
- SelectInfer (arXiv 2026)
- MoRAGBench