

SLOAI x sference: What a Trillion Tokens Taught Us About Inference
Open-weight models are easy to run and hard to run well. So where do latency and cost actually come from once real traffic hits?
Join us as SLOAI hosts Jernej Štrasner, co-founder and CEO of sference, for an evening on the infrastructure side of building with AI. With a trillion tokens served behind them, Jernej will give a no-fluff look at where latency and cost come from in production LLM serving, what moved the needle for the sference team and what didn't, and what it means for how you build agents and apps on top.
Built for AI engineers, infrastructure people, and founders shipping products on LLMs. The talk is in English. Expect practical lessons, an open Q&A, and time to meet the community afterwards.
About the speaker
Jernej Štrasner is the co-founder and CEO of sference, a European AI compute platform serving open-weight models. He previously founded Specto (acquired by Sentry), led infrastructure engineering at Sentry, and worked at Meta. He spent seven years in San Francisco and now builds from Slovenia.
When & where
📅 Tuesday, 1 December 2026
📍 Ljubljana.tech, Trg mladih 9, 1000 Ljubljana
Agenda
17:50 - Doors open
18:00 - Welcome from SLOAI
18:10 - Talk by Jernej Štrasner
18:50 - Q&A
19:10 - Networking
Organised by SLOAI, featuring sference, and hosted at Ljubljana.tech.