Call For Papers: Special Issue on Artificial Intelligence and Extreme-Scale Workflows

Important Dates

• Submissions due: 1 February 2027
• Publication date: July - September 2027

As AI transforms scientific discovery, a powerful convergence is underway between artificial intelligence and extreme-scale workflows. These workflows are the systems that orchestrate complex, multi-step scientific campaigns on high-performance computing (HPC) infrastructure. This convergence gives rise to AI workflows, encompassing two complementary directions:

• Workflows for AI: extreme-scale workflows that integrate AI software and hardware components for scientific and engineering tasks (e.g. training, inference, surrogate modeling, uncertainty quantification, etc.) alongside traditional HPC simulations and implementations.

• AI for workflows: using AI techniques and tools to optimize workflow execution itself, including AI-driven scheduling, development, autonomous steering, and self-healing fault recovery.

While the opportunities are substantial, significant challenges remain. AI tasks exhibit fundamentally different behaviors from traditional HPC workloads such as distinct I/O patterns, software stacks, and requirements for asynchronous execution and long-running services that existing workflow management systems and HPC systems were not designed to support. Meanwhile, the rise of agentic AI and large language models introduces new possibilities for workflow automation that the community has only begun to explore.

This special issue builds on the success of the International Symposium on Artificial Intelligence and Extreme-Scale Workflows (AIExScale), co-located with SC25. The inaugural edition featured a keynote by Ian Foster and invited talks by Debbie Bard, Rafael Ferreira da Silva, and Katrin Heitmann, attracting close to 100 attendees.

We invite high-quality articles that advance the understanding, design, implementation, and application of AI workflows, and that bridge the historically distinct HPC and AI communities. We welcome submissions from workflow system developers, AI/ML researchers, HPC systems architects, and domain scientists. Manuscripts describing cross-community collaborations are particularly encouraged.

Important: Submissions must center on how AI advances scientific workflows, or how workflows leverage AI to advance specific science domains; contributions whose core advance is a model or ML algorithm without a workflow dimension are outside the scope of this issue.

Topics of interest include, but are not limited to:

• AI-integrated scientific workflows: Design and deployment of workflows incorporating AI training, inference, surrogate modeling, and uncertainty quantification alongside HPC simulations.
• AI-driven workflow optimization: Machine learning techniques for workflow scheduling, resource allocation, and adaptive execution.
• Autonomous and self-healing workflows: AI-enabled autonomous steering, fault detection, and self-recovery in large-scale scientific campaigns.
• Agentic and LLM-driven workflows: Applications of large language models and agentic AI systems for scientific workflow planning, automation, and orchestration.
• HPC–AI software stack convergence: Strategies for bridging MPI-based simulation codes and Python-based ML frameworks within unified workflow environments.
• I/O and data management for AI workflows: Approaches to managing the distinct I/O patterns, streaming data, and long-running services (vector databases, message brokers) required by AI components on HPC systems.
• Workflow management systems for AI workflows: Advances in WMS to support asynchronous execution, on-demand task spawning, and heterogeneous resource management.
• Domain science case studies: Real-world AI workflow applications in fields such as molecular dynamics, materials science, cosmology, biochemistry, and climate science, with generalizable lessons.
• Benchmarking, reproducibility, and evaluation: Frameworks and methodologies for assessing the performance, scalability, and reproducibility of AI workflows.

Submission Guidelines

For author information and guidelines on submission criteria, visit the Author’s Information page. Articles should be between 2,400 and 6,250 words, including all main body, abstract, keyword, bibliography (25 references or less), and biography text. Each table and figure counts for 250 words.

Please submit papers through the IEEE Author Portal and be sure to select the special issue or special section name. Manuscripts should not be published or currently submitted for publication elsewhere. Please submit only full papers intended for review, not abstracts. If requested, abstracts should be sent by email to the guest editors directly.

Guest Editors

• Loic Pottier, Lawrence Livermore National Laboratory, USA
• Orcun Yildiz, Argonne National Laboratory, USA
• Nargess Memarsadeghi, NASA’s Goddard Space Flight Center , USA