DGX Spark & Sovereign AI: Architecting Air-Gapped Agents
The shift from mainframes to PCs in the 1980s removed gatekeepers by putting computing power directly in the hands of engineers. Today, NVIDIA DGX Spark does the same for AI, giving builders ownership of the machine instead of leaving compute, timelines, and progress under institutional control.
In this session, I’ll show you how to engineer a Sovereign AI Lab that decouples the speed of development from the scale of production. We’ll cover: - The Offline Iteration Loop: Bring raw compute to your desk and debug agentic systems end-to-end without triggering a single security review. - Data Sovereignty for Speed: Keeping sensitive data strictly localized isn’t just a compliance strategy, it’s the fastest path to production. - The Lab-to-Cloud Handoff: Local is where you build fast. The enterprise cloud is where you scale. I will show you the architectural bridge between the two.
Walk away with a reference architecture to secure your data, reclaim your iteration speed, and ship Agentic AI.
Thiago Shimada Ramos
Thiago Shimada Ramos is a NVIDIA-Certified Professional in Agentic AI. As one of the few experts globally to hold the “Elite Trifecta” of CNCF Kubestronaut, Docker Captain, and Microsoft MVP, he specialises in helping regulated enterprises bridge the gap between AI innovation and strict data compliance. Based in Melbourne, where he leads the local Cloud Native and Docker communities, Thiago’s current work focuses on architecting Sovereign AI Labs, an air-gapped infrastructure that decouples the speed of localized AI development from the scale of enterprise production.