From Automation to Autonomy: Agentic AI in HPC • Fawada Qaiser
High-performance computing has automated many aspects of scientific research, yet researchers still spend substantial time coordinating workflows across simulations, data, models, schedulers, and analysis tools.
This talk explores how agentic AI could move HPC from automation to autonomy by enabling workflows that reason, plan, act, and adapt across multi-step scientific tasks. It highlights opportunities in simulation steering, adaptive exploration, and hybrid simulation-AI pipelines.
The central message is that agentic AI is not replacing scientists, but reducing workflow complexity and accelerating discovery.
Panel Discussion: The Future of AI in Scientific Computing
Moderator: Dr Fawada Qaiser
*Panel Members:* - Prof Jeyan Thiyagalingam - Prof Mark Wilkinson - Daniele Sorini - Gavin Leroy - Craig Bower
This panel discusses emerging directions in AI-enabled scientific computing, including trustworthy AI, digital twins, agentic workflows, and the role of HPC in future discovery.