01
From LLM Inference to Agentic Workloads: Characterization and Implications for Serving Systems
arXiv:2608.15127 · 2026
I build systems for efficient, reliable agentic workloads and large language model serving. My current work spans serving systems, sandboxing and checkpoint/restore, inference pipelines, and the systems foundations needed to run AI applications at scale.
Agent Serving Agent Sandboxes LLM Serving MLSys ML Inference Pipelines ML Compilers Distributed DNN Training Video Analytics Systems