Johnny Núñez Cano

Johnny Núñez Cano

AI Developer Advocate at NVIDIA — Physical AI, Robotics & Edge Computing. Applied Research & Solutions Architect. Championing open-source models, frameworks, and hardware. CVPR 2026 Tutorial Co-organizer. PhD in Computer Vision.

edge2cloud — ML Containers for NVIDIA CUDA SBSA

Published:

edge2cloud is a collection of Machine Learning containers for NVIDIA CUDA SBSA (Server Base System Architecture) spanning Jetson, DGX Spark, and cloud deployments.

Motivation

The NVIDIA ecosystem now spans a unified architecture from edge (Jetson Orin, Jetson Thor) through workstation (DGX Spark) to cloud (DGX, HGX). edge2cloud provides a unified containerized ML stack that works seamlessly across all these platforms, enabling developers to build once and deploy everywhere.

Features

  • Unified CUDA SBSA container images for Jetson, DGX Spark, and cloud
  • Pre-built ML frameworks (PyTorch, TensorFlow, ONNX Runtime) optimized per platform
  • Supports CUDA 13.2 and latest JetPack versions
  • CI/CD-ready for fleet management and continuous deployment workflows
  • Bridges the gap between edge prototyping and production cloud deployment

Architecture

Edge (Jetson Orin/Thor) → Workstation (DGX Spark) → Cloud (DGX/HGX)
         └──────────── Unified SBSA Container Stack ────────────┘