BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.kiwipycon.nz//kiwi-pycon-xiv-2025//speaker//HRNV7
 T
BEGIN:VTIMEZONE
TZID:Pacific/Auckland
BEGIN:DAYLIGHT
DTSTART:20241122T000000
TZNAME:NZDT
TZOFFSETFROM:+1300
TZOFFSETTO:+1300
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20250406T030000
RDATE:20260405T030000
TZNAME:NZST
TZOFFSETFROM:+1300
TZOFFSETTO:+1200
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20250928T030000
RDATE:20260927T030000
TZNAME:NZDT
TZOFFSETFROM:+1200
TZOFFSETTO:+1300
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:Unlocking Document Intelligence with Open-Source AI - Mingxuan Zha
 o
DTSTART;TZID=Pacific/Auckland:20251122T141500
DTEND;TZID=Pacific/Auckland:20251122T144500
DTSTAMP:20260815T124211Z
UID:pretalx-kiwi-pycon-xiv-2025-J7NBXS@talks.kiwipycon.nz
DESCRIPTION:Most organizational knowledge is still locked inside complex d
 ocuments\, making it difficult to extract and use the information effectiv
 ely. Traditional tools often fail when working with real-world PDFs. Table
 s lose their structure\, figures are separated from captions\, and multi-c
 olumn layouts are flattened into unreadable text. These issues create a si
 gnificant barrier to using AI on real document data.\n\nThe open-source pr
 oject Docling presents a new approach to document ingestion that mirrors h
 uman comprehension using open-source deep learning models in a neat Python
  package. The system extracts structured information through consistent AP
 Is\, preserving original document hierarchy while ensuring machine readabi
 lity.\n\nWith support for over ten of the most common file formats and a c
 onsistent API\, Docling enables production-ready document processing pipel
 ines and provides seamless integration with established frameworks includi
 ng LangChain and LlamaIndex\, as well as multilingual support. Its MIT lic
 ense and local execution model make it suitable for sensitive enterprise a
 pplications.
LOCATION:Breakout Room
URL:https://talks.kiwipycon.nz/kiwi-pycon-xiv-2025/talk/J7NBXS/
END:VEVENT
END:VCALENDAR