~/talk

Frederic Van Haren | Supercloud 4

Talkat Supercloud 4 · 2023-10-24..10-25· [cloud-infrastructure]
links:none on file
slides:none on file

Argues that AI decisions hinge on infrastructure choices not buzzwords: training often stays on-prem training, cloud inference splits because GPUs cost and inference is pay-as-you-go; many teams buy GPUs without redesigning servers or networks and then hit latency and compatibility failures. Recommends transfer learning rather than building huge models and points to Hugging Face models, edge filtering for video ingestion, and a CFO use case building an on-prem LLM for SEC reporting to keep data lineage and security.

AI-written summary · about

  1. 0:00Introduction
  2. 1:03What has been happening in AI
  3. 1:57Mistakes made in AI
  4. 4:12Data quality challenges
  5. 5:32Too much data
  6. 6:46Data lineage
  7. 8:12Training and inference
  8. 10:58Why use AI
  9. 12:34What AI cant do
  10. 14:43Transfer learning
  11. 16:21Data management
~/search
type to search 6,518 conferences, 143,918 speakers and 192,133 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
← →
on a talk: 5 seconds back / forward
0–9
on a talk: jump to 0%–90% of the recording
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.