NVIDIA DGX Cloud

DGX Cloud is the environment NVIDIA builds AI in: open-source frontier and foundational models are trained on it, new system architectures are validated on it, and production AI workloads run on it — across NVIDIA's own infrastructure and its cloud partners. It is also the proving ground. Problems that only appear at that scale get solved there first, and what comes out becomes the software and reference architectures the rest of the ecosystem inherits.

I lead the team responsible for its user experience.

The interesting part of that job is the gap between the scale of the system and the size of the screen. A job here can hold thousands of GPUs for weeks, and something in a fleet that size is always degraded. So the work is less about screens and more about which few facts, out of an overwhelming number of true ones, an engineer needs in front of them to decide what to do next.

I came to this through Lepton AI, which NVIDIA acquired in 2025.