A 3-hour technical deep dive · Florin Bădiță
SFO 09:00 PDT · NYC 12:00 EDT · London 17:00 BST · Bucharest 19:00 EEST
This is a short teaser. The full 50-slide deck goes to registered attendees after the session.
Every token an LLM produces, it looks at every other token. That's the secret of the transformer — and the reason context length matters so much.
In the live session we trace one token end-to-end: embedding → attention → softmax → prediction.
Names you've heard. Live, you'll see when each one actually works.
The misconceptions that quietly cost people months.
Friday 22 May 2026
SFO 09:00 PDT · NYC 12:00 EDT · London 17:00 BST · Bucharest 19:00 EEST
Questions? florin@badita.org