BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//frankwinkler.me//speaking//EN
NAME:Frank Winkler — Speaking
X-WR-CALNAME:Frank Winkler — Speaking
BEGIN:VEVENT
UID:keynote-applied-genai-2026@frankwinkler.me
SEQUENCE:0
DTSTAMP:20260621T153138Z
DTSTART;VALUE=DATE:20260918
X-MICROSOFT-CDO-ALLDAYEVENT:TRUE
X-MICROSOFT-MSNCALENDAR-ALLDAYEVENT:TRUE
SUMMARY:[ Talk title — keynote on applied generative AI ] — [ Conferen
 ce name ]
LOCATION:[ City ]
DESCRIPTION:A placeholder abstract: where generative AI is genuinely chang
 ing how teams build\, and where it's still theatre — with a framework fo
 r telling the two apart.
END:VEVENT
BEGIN:VEVENT
UID:workshop-mlops-aws-2026@frankwinkler.me
SEQUENCE:0
DTSTAMP:20260621T153138Z
DTSTART;VALUE=DATE:20260730
X-MICROSOFT-CDO-ALLDAYEVENT:TRUE
X-MICROSOFT-MSNCALENDAR-ALLDAYEVENT:TRUE
SUMMARY:[ Workshop title — hands-on MLOps on AWS ] — [ Meetup / event 
 ]
LOCATION:Singapore
DESCRIPTION:A placeholder abstract for a half-day\, build-along workshop. 
 Bring a laptop\; leave with a deployed\, evaluated pipeline.
END:VEVENT
BEGIN:VEVENT
UID:trustworthy-ai-products-2026@frankwinkler.me
SEQUENCE:0
DTSTAMP:20260621T153138Z
DTSTART;VALUE=DATE:20260402
X-MICROSOFT-CDO-ALLDAYEVENT:TRUE
X-MICROSOFT-MSNCALENDAR-ALLDAYEVENT:TRUE
SUMMARY:Designing trustworthy AI products: from model output to calibrated
  interface — AWS Community Day
LOCATION:Singapore
DESCRIPTION:A hands-on\, half-day session pairing cloud architecture with 
 product design. We took a working RAG model and built the trust layer arou
 nd it — confidence signals\, citations\, and failure states — then mea
 sured how each change moved users from over-trust to genuine calibration.
END:VEVENT
BEGIN:VEVENT
UID:genai-adoption-apac-2026@frankwinkler.me
SEQUENCE:0
DTSTAMP:20260621T153138Z
DTSTART;VALUE=DATE:20260214
X-MICROSOFT-CDO-ALLDAYEVENT:TRUE
X-MICROSOFT-MSNCALENDAR-ALLDAYEVENT:TRUE
SUMMARY:The state of GenAI adoption in APAC: shipping vs. piloting — Tec
 h in Asia Conference
LOCATION:Jakarta
DESCRIPTION:A four-person panel cutting through the hype cycle: where gene
 rative AI is in real production across the region\, where it's stuck in pr
 oof-of-concept purgatory\, and the organisational — not technical — re
 asons why. I argued that the bottleneck is rarely the model.
END:VEVENT
BEGIN:VEVENT
UID:model-to-product-devfest-2025@frankwinkler.me
SEQUENCE:0
DTSTAMP:20260621T153138Z
DTSTART;VALUE=DATE:20251129
X-MICROSOFT-CDO-ALLDAYEVENT:TRUE
X-MICROSOFT-MSNCALENDAR-ALLDAYEVENT:TRUE
SUMMARY:From model to product: the unglamorous work that earns trust — D
 evFest SG
LOCATION:Singapore
DESCRIPTION:The opening keynote on everything that happens after the eval 
 numbers look good — latency as a trust signal\, the design of failure st
 ates\, and the feedback loops that compound. The talk that later became th
 e most-read piece on the blog.
END:VEVENT
END:VCALENDAR