BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//cfp.pydata.org//pydata-eindhoven-2025//talk//7FTJYA
BEGIN:VEVENT
UID:pretalx-pydata-eindhoven-2025-7FTJYA@cfp.pydata.org
DTSTART:20251209T152000Z
DTEND:20251209T155000Z
DESCRIPTION:Traditional data pipelines often tie ingestion and transformati
 on together\, forcing data into rows and columns early in the process. But
  modern workloads - from large XML documents to high-resolution video file
 s - have transformations that are far from trivial and require very differ
 ent compute resources than ingestion. Separating these concerns becomes es
 sential. When a transform fails\, you shouldn’t have to re-download data
  or hit the source system again.
DTSTAMP:20260425T161905Z
LOCATION:Planck-Bohr
SUMMARY:Ingestify: Rethinking Ingestion for Complex Data - Koen Vossen
URL:https://cfp.pydata.org/pydata-eindhoven-2025/talk/7FTJYA/
END:VEVENT
END:VCALENDAR