AI agents & workflows · 2026

Overcoming Orchestration Bottlenecks at Exascale: A Decentralized, Policy-Driven Approach for Sim-AI Ensembles

Harikrishna Tummalapalli, Christine M. Simpson, Riccardo Balin, Vitali A. Morozov, Thang D. Pham, Murat Keçeli, Thomas D. Uram

arXiv:2607.12211 (2026)

preprint arXiv PDF

Summary

EnsembleLauncher is a recursively hierarchical, fully decentralized workflow orchestrator with a programmable scheduling-policy interface that scales heterogeneous simulation–AI ensembles on the Aurora exascale system.

Keywords

Cite

@article{2026_tummalapalli_ensemblelauncher,
  title = {Overcoming Orchestration Bottlenecks at Exascale: A Decentralized, Policy-Driven Approach for Sim-AI Ensembles},
  author = {Harikrishna Tummalapalli and Christine M. Simpson and Riccardo Balin and Vitali A. Morozov and Thang D. Pham and Murat Keçeli and Thomas D. Uram},
  journal = {arXiv:2607.12211 (2026)},
  year = {2026},
  eprint = {2607.12211},
  archivePrefix = {arXiv}
}