BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260731T133317Z
LOCATION:Bldg. 8 - B 102
DTSTART;TZID=Europe/Stockholm:20260701T140000
DTEND;TZID=Europe/Stockholm:20260701T160000
UID:submissions.pasc-conference.org_PASC26_sess165@linklings.com
SUMMARY:MS5H - Co-Design of Workflow Orchestration Systems for AI/ML-Enhan
 ced Scientific Workloads on HPC Infrastructure
DESCRIPTION:Access the recording\n\nOrganizer(s): Ozgur Ozan Kilic (Brookh
 aven National Laboratory), and Charles Leggett (Lawrence Berkeley National
  Laboratory)\n\nScientific workflows increasingly combine traditional simu
 lations with AI/ML methods, creating complex pipelines that must operate a
 cross heterogeneous, distributed infrastructure. This minisymposium examin
 es how co-design approaches—bringing together domain scientists, HPC archi
 tects, and AI researchers—can address the resulting orchestration challeng
 es. We explore three interconnected themes. First, we examine intelligent 
 workflow generation, including how agentic AI and large language models ca
 n assist scientists in developing and optimizing workflows for heterogeneo
 us architectures. Second, we address federated infrastructure services, di
 scussing architectural patterns for unified APIs that enable workflows to 
 seamlessly traverse institutional boundaries while respecting security and
  data governance constraints. Third, we consider AI-ready data and model s
 ervices, exploring how federated catalogs, automated curation, and model r
 egistries can accelerate discovery. While drawing examples from High Energ
 y Physics—where petabyte-scale data and international collaborations have 
 driven workflow innovation—the patterns we discuss apply broadly to climat
 e science, materials research, genomics, and other data-intensive domains.
  Our concluding panel examines future directions for co-designed scientifi
 c computing systems.\n\nLessons in Scale: Exploring the Potential of AI Ag
 ents for Global Compute Orchestration\n\nLarge-scale High Energy Physics (
 HEP) experiments, such as ATLAS at the LHC, have spent decades mastering t
 he art of the "heterogeneous ensemble"—knitting together geographically di
 stributed grid sites, Kubernetes clusters, HPC centers, and commercial clo
 uds. While these systems have successfu...\n\n\nKaushik De (University of 
 Texas at Arlington)\n---------------------\nFlowgentic: An Execution Coord
 ination Layer for Adaptive, Agent-Driven HPC Workflows\n\nModern scientifi
 c workflows increasingly combine large-scale HPC simulations with machine 
 learning inference, adaptive decision-making, and dynamic control flow. Wh
 ile agentic orchestration frameworks like LangGraph provide expressive abs
 tractions for goal-directed coordination, they assume low-late...\n\n\nMat
 teo Turilli (IE University)\n---------------------\nCo-Designed AI Service
 s for HEP Software Understanding and Portability\n\nAI-ready scientific wo
 rkflows require more than data pipelines and model endpoints; they also re
 quire software stacks that domain scientists can understand, maintain, and
  adapt for heterogeneous HPC systems. This talk presents CelloAI, a locall
 y hosted AI assistant for High Energy Physics software d...\n\n\nMohammad 
 Atif (Brookhaven National Laboratory)\n---------------------\nFuture Direc
 tions for Co-Designed Scientific Computing\n\nThis panel brings together p
 erspectives from across the minisymposium to examine how co-design methodo
 logies can evolve alongside scientific requirements and emerging computing
  paradigms, including AI-driven workflow orchestration and heterogeneous H
 PC infrastructures. Panelists will discuss governa...\n\n\nOzgur Ozan Kili
 c (Brookhaven National Laboratory)\n\nDomain: Engineering, Physics, Comput
 ational Methods and Applied Mathematics\n\nSession Chair: Ozgur Ozan Kilic
  (Brookhaven National Laboratory)
END:VEVENT
END:VCALENDAR