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:20260624T171341Z
LOCATION:Bldg. 8 - B 102
DTSTART;TZID=Europe/Stockholm:20260701T150000
DTEND;TZID=Europe/Stockholm:20260701T153000
UID:submissions.pasc-conference.org_PASC26_sess165_msa225@linklings.com
SUMMARY:Co-Designed AI Services for HEP Software Understanding and Portabi
 lity
DESCRIPTION:Mohammad Atif (Brookhaven National Laboratory)\n\nAI-ready sci
 entific workflows require more than data pipelines and model endpoints; th
 ey also require software stacks that domain scientists can understand, mai
 ntain, and adapt for heterogeneous HPC systems. This talk presents CelloAI
 , a locally hosted AI assistant for High Energy Physics software developme
 nt that combines large language models with retrieval-augmented generation
  to support documentation, code comprehension, and dependency-aware code g
 eneration. CelloAI produces Doxygen-style comments and file summaries from
  project-specific sources, while its syntax-aware retrieval and call-graph
  integration help preserve context during refactoring and optimization. Ev
 aluations on representative software from ATLAS, CMS, and DUNE illustrate 
 how these capabilities can reduce the effort needed to port and evolve leg
 acy applications for modern architectures. The talk argues that trustworth
 y AI services for science must be co-designed with domain scientists, soft
 ware engineers, and facility operators: local deployment supports privacy 
 and operational control, retrieval-based workflows improve transparency, a
 nd experiment-specific knowledge bases keep generation grounded in real sc
 ientific code. In this sense, CelloAI complements data and model services 
 by providing an AI-ready software knowledge layer for next-generation HEP 
 computing.\nThis version is also based on the paper’s description of the s
 ystem and its use cases.\n\nDomain: Engineering, Physics, Computational Me
 thods and Applied Mathematics\n\nSession Chair: Ozgur Ozan Kilic (Brookhav
 en National Laboratory)\n\n
END:VEVENT
END:VCALENDAR