Sovereign AI and interactive HPC: unifying training, inference, and exploration in one workflow

What you'll learn

• Why most organizations run AI training and inference on separate platforms—and the operational cost of that split
• How service endpoints unify batch computing (training) and persistent services (inference) in single workflow definitions
• The architecture for sovereign AI: fine-tuning and serving models entirely on your infrastructure
• Live demo: training → inference → interactive testing in one workflow, with no external data exposure
• How to use service endpoints for real-time workflow interaction

Jonathon Anderson

Principal HPC Product Engineer, CIQ
Jonathon leads HPC product development at CIQ, with deep expertise in workflow orchestration, container technologies, and high-performance computing architecture. He has worked with national laboratories, research institutions, and enterprises to design HPC systems that balance performance with usability.

David Godlove

Senior HPC Engineer, CIQ
Dave Godlove is the technical product writer for the Fuzzball team, acting as a vital link between customers and CIQ engineers. With a background as an HPC user, staff scientist, and container developer, he brings a deep understanding of the challenges faced by HPC end users.
Trained in neuroscience at Vanderbilt University, Dave completed his postdoctoral work at the NIH. It was there that he discovered his passion for HPC and joined the NIH’s Biowulf cluster as a staff scientist. As a founding member of the HPC container community, he has devoted nearly a decade helping guide development of Apptainer and educating users on container-based HPC workflows.