AgenticAI4HPC 2026

Introduction

Recent progress in agentic AI—AI systems capable of autonomous planning, reasoning, adaptation, and tool use—opens transformative opportunities across the HPC research landscape. Beyond system-level optimization, agentic AI can play a pivotal role in code modernization, automated code translation across programming models, performance tuning for diverse hardware platforms, and the long-term sustainability of large scientific software stacks.

As HPC systems continue to evolve toward exascale and beyond, developers face increasing challenges in maintaining legacy codes, porting applications to emerging architectures, and achieving performance portability across changing programming models and runtime environments. Agentic approaches offer the potential to autonomously refactor legacy codes, translate between models, recommend architecture-aware optimizations, and iteratively tune applications.

AgenticAI4HPC’26 will bring together researchers and practitioners to explore methodologies, systems, and real-world applications that leverage agentic AI to accelerate HPC software development, enhance productivity, and enable sustainable performance at scale. By fostering interdisciplinary dialogue at the intersection of autonomous AI systems and advanced computing, the workshop aims to define the future role of agentic AI in advancing the entire HPC research ecosystem.

Objectives, Scope and Topics

The main goal of this workshop is to build bridges between HPC and AI by fostering interdisciplinary dialogue at the intersection of autonomous AI systems and advanced computing, and to define the future role of agentic AI in advancing the entire HPC research ecosystem. Topics of interest include, but are not limited to:

Programming Models, Code Modernization & Sustainability
- Agent-based support for HPC programming models
- Autonomous code translation across programming models and hardware backends
- Intelligent runtime systems and adaptive execution frameworks
- Automated modernization of legacy HPC applications
- AI-driven refactoring for scalability, maintainability, and portability
- Sustainable software engineering practices enabled by agentic systems
Scientific Workflows & Application Development
- Intelligent workflow composition and orchestration
- Iterative optimization through reinforcement learning or multi-agent systems
- Agent-assisted debugging, verification, and validation
- AI-driven parallel and distributed algorithm design
- Integration of simulation, data analytics, and AI workloads
Co-Design & Human–AI Collaboration in HPC
- Human-in-the-loop agentic systems for scientific computing
- Hardware–software–application co-design with autonomous agents
- Productivity tools and AI copilots for HPC developers
- Trust, transparency, and reproducibility in agent-driven HPC systems
Evaluation, Benchmarking & Reliability
- Benchmarking agentic AI approaches in HPC environments
- Reliability, robustness, and fault tolerance of autonomous systems
- Measurement methodologies for productivity and performance gains
- Case studies and real-world deployments of agentic AI in HPC

Program (Tentative)

9:00 – 9:05 Opening Remarks

9:05 – 9:40 Invited Talk 1 (Ali Jannesari): From Code Generation to Performance Reasoning — Agentic AI for the Autonomous HPC Stack

9:40 – 10:00 Paper Presentation 1

10:00 – 10:30 Morning break

10:30 – 11:05 Invited Talk 2 (Abhinav Bhatele): Agentic Approaches to Generate and Optimize Parallel Code

11:05 – 12:20 Paper Presentations 2

12:20 – 12:30 Best Paper Recognition and Closing

Important Dates

Submission August 1, 2026 (AoE)
Acceptance September 4, 2026 (AoE)
Camera Ready September 20, 2026 (AoE)
Workshop During SC 2026

Invited Talks

Invited Talk 1 · Confirmed

From Code Generation to Performance Reasoning: Agentic AI for the Autonomous HPC Stack

Ali Jannesari — Associate Professor, Iowa State University, USA

Ali Jannesari is an Associate Professor and Director of the Laboratory for Software Analytics and Pervasive Parallelism in the Department of Computer Science at Iowa State University. His research focuses on the intersection of High-Performance Computing (HPC) and Artificial Intelligence (AI). Dr. Jannesari has published over a hundred refereed articles, several of which have received awards.

Invited Talk 2 · Confirmed

Agentic Approaches to Generate and Optimize Parallel Code

Abhinav Bhatele — Associate Professor, University of Maryland, USA

Abhinav Bhatele is an Associate Professor in the Department of Computer Science and Director of the Parallel Software and Systems Group at the University of Maryland, College Park. His research interests are broadly in systems and AI, with a focus on parallel computing and distributed AI. He has published research in parallel programming models and runtimes, network design and simulation, applications of machine learning to parallel systems, parallel deep learning, and on analyzing, visualizing, modeling and optimizing the performance of parallel software and systems.

Organizers (Contact us)

Mohammad Alaul Haque Monil (Chair) Oak Ridge National Laboratory, USA monilm@ornl.gov

Pedro Valero-Lara (Co-Chair) Oak Ridge National Laboratory, USA valerolarap@ornl.gov

Daichi Mukunoki (Co-Chair) Nagoya University, Japan mukunoki.daichi.p2@f.mail.nagoya-u.ac.jp

Bogdan Nicolae (Co-Chair) Argonne National Laboratory, USA bnicolae@anl.gov

Steering Committee

Jeffrey S. Vetter Section Head, Oak Ridge National Laboratory, USA

Sunita Chandrasekaran Associate Professor & Co-Director, AI Center of Excellence, University of Delaware, USA

Antonio J. Peña Senior Researcher, Barcelona Supercomputing Center, Spain

Ignacio Laguna Computer Scientist, Lawrence Livermore National Laboratory, USA

Toshiyuki Imamura Team Leader, RIKEN Center for Computational Science, Japan

Franz Franchetti Professor, Carnegie Mellon University, USA

Abhinav Bhatele Associate Professor, University of Maryland, USA

Ali Jannesari Associate Professor, Iowa State University, USA

Rio Yokota Professor, Institute of Science Tokyo, Japan

Programme Committee

Note: The list below shows the currently confirmed members. Additional members are being added — please check back soon.

Upasana Sridhar

Carnegie Mellon University, USA

Johannes Doerfert

Lawrence Livermore National Laboratory, USA

Xingfu Wu

Argonne National Laboratory, USA

Tanzima Islam

Texas State University, USA

Olivier Aumage

INRIA, France

William F. Godoy

Oak Ridge National Laboratory, USA

Simon Garcia De Gonzalo

Sandia National Laboratory, USA

Aaron Young

Oak Ridge National Laboratory, USA

Diego Andrade Canosa

University of A Coruña, Spain

Eduardo Iraola

Barcelona Supercomputing Center, Spain

Grace Mclewee

University of Oregon, USA

Arijit Bhattacharjee

Iowa State University, USA

Manuscript Submission

We invite submissions of original, unpublished research and experiential papers. Full papers should be up to 10 pages in length, formatted in the standard IEEE conference format including references. All paper submissions will be managed electronically via the SC26 workshop submission system. NOT YET OPEN FOR SUBMISSION.

Proceedings

Accepted papers will be published in the IEEE Digital Library as part of the workshop proceedings of SC 2026.

Best Paper Recognition

The Best Paper Recognition will be selected on the basis of explicit recommendations of the reviewers and their scoring towards the paper’s originality and quality.

Reproducibility

Submitted papers must meet the requirements of the SC26 reproducibility initiative, including Artifact Description (AD) and Artifact Evaluation (AE). Paper submissions are single-blinded; reviewers will not be known to the authors. The technical committee will be instructed to strive toward an ethical and transparent peer-review process, including declaring conflicts of interest with the authors. Each paper will receive at least three independent reviews. Acceptance criteria will be determined solely on the technical merits of the work, its impact on the accessibility of the presented paper, and the quality of the final manuscript.

Registration

Information about registration is available at the SC 2026 website.