Workshops Archive

Benchmarking Agentic AI Workflows with Charliecloud


Workshop: 8th International Workshop on Containers and New Orchestration Paradigms for Isolated Environments in HPC (CANOPIE-HPC)

Authors: Silba Dowell (Los Alamos National Laboratory)

Abstract: Agentic AI workflows are an emerging paradigm in the AI space that enable autonomous agents to drive the traditional prompt token response through multi-step reasoning, tool calling, and planning phases. Recent advances to orchestrate production level frameworks such as these are trending towards using LLM inference engines such as vLLM and model abstraction layers like LiteLLM. Charliecloud enhances this work by leveraging portability across diverse HPC systems. In this talk, we address why such frameworks are leveraging these tools with Charliecloud and how we can assess end-to-end performance of these workflows through benchmarking.


Back to 8th International Workshop on Containers and New Orchestration Paradigms for Isolated Environments in HPC (CANOPIE-HPC) Archive Listing

Back to Full Workshop Archive Listing