Where Intelligence Can Live: The Full Power of Local AI, Live on Stage

Ivan Fioravanti runs generative AI entirely on hardware he owns, with no cloud calls, and does it live on stage. The demos run on a small rack of Mac Studio, MacBook Pro and NVIDIA DGX Spark machines generating text, code, images, speech, music and video. He'll cover memory limits, quantisation trade-offs, splitting big models across machines, Apple Silicon versus NVIDIA, and the cost case for moving batch work on-premises. Fioravanti co-founded the Microsoft integrator 4ward in 2001 and CoreView in 2014, and stepped back from operations this year to focus on local AI, including work with Apple's MLX team. Hosted by Kaggle & Practical AI Milano at OverNet Education in Milanofiori, Assago. Doors 18:15, talk 18:45, Q&A and networking until 20:30. Free, but in-person places are capped and Eventbrite currently shows it sold out. A YouTube stream is planned but not guaranteed. Why go: local inference benchmarked live by someone who has actually moved spend off the cloud. Best for ML engineers and CTOs weighing on-prem against API costs.