BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Optimizing Data Delivery and Scalable HI Profile Classification fo
 r the SKA Era: Infrastructure and Science Challenges at the Spanish SRC - 
 Manuel Parra-Royon
DTSTART:20250701T101500Z
DTEND:20250701T110000Z
UID:TALK234028@talks.cam.ac.uk
CONTACT:Dr. Manu Parra-Royón
DESCRIPTION:This talk presents ongoing work at the Spanish SKA Regional Ce
 ntre (esSRC) in the context of the SRCNet 0.1. The first part focuses on t
 he development of efficient data delivery techniques from the distributed 
 Rucio-based storage system to the SRC infrastructure and\, ultimately\, to
  user workspaces. Several approaches have been evaluated to support scienc
 e-ready access\, yet current solutions often involve unnecessary data dupl
 ication in user areas\, resulting in increased usage of storage and comput
 ational resources. To address this\, we have prototyped mechanisms based o
 n file linking\, caching\, and data reuse\, enabling more efficient access
  paths for users. While these methods show promising improvements in terms
  of performance and resource usage\, challenges remain\, particularly in t
 erms of orchestration\, scalability\, and compatibility with existing work
 load managers.\n\nThe second part presents advances in the automated class
 ification of neutral hydrogen (HI) profiles using machine learning methods
 \, building on previous work [Parra et al.\, 2024\, arXiv:2501.11657]. We 
 outline a roadmap for extending these techniques to handle the data volume
 s expected from the SKA Observatory. This includes developing scalable pip
 elines capable of ingesting and processing large spectral datasets in a re
 producible and efficient manner\, and adapting the classification models t
 o cope with the diversity and complexity of the SKA data products. 
LOCATION:Venue to be confirmed
END:VEVENT
END:VCALENDAR