Liting Hu Research Group

University of California, Santa Cruz

How can edge and cloud systems provide low-latency, high-throughput stream and AI services across distributed infrastructures? The ELVES Research Lab implements adaptive edge stream processing projects such as DART and FP4S to scale analytics engines across Spark, Flink, and Storm. Researchers design federated learning and AI-serving platforms (Totoro, Totoro+) using distributed runtime mechanisms to train and serve models at the edge. The group develops lightweight virtualization and serverless scheduling systems (Capybara, Ekko) using container and decentralized scheduling techniques for elastic compute. They build intelligent resource orchestration control planes (RBay, Monalytics) applying monitoring-driven control for edge and datacenter clusters to optimize utilization.