NVIDIA GPU Accelerated Application Samples on Microsoft Azure

Table of Contents

• Description
• Support Level
• Requirements
• Quickstart
• Samples
• Usage
• Known Issues
• Contributions
• Support
• License

Description

This repository maintains sample applications designed for NVIDIA software tools integrated with Microsoft Azure.

For select demonstrations, the sample code is contained within this repository. For others, we reference and link to exceptional demonstrations available outside of this repository.

Support Level

These samples are provided as community examples and are not covered by NVIDIA Enterprise Support. They are intended as reference implementations to help users get started with NVIDIA software on Azure. See Support below for how to get help.

Requirements

• An active Microsoft Azure account with permissions to create the resources used by a given sample
• Access to NVIDIA GPU-enabled Azure VM sizes in your target Azure region (for example, A10, A100, H100, or H200 GPU families)
• The Azure CLI installed and configured, or access to the Azure portal
• An NGC API key for NVIDIA software access
• Additional, sample-specific prerequisites are documented in each sample's own README

Quickstart

1. Clone this repository:

   git clone https://github.com/NVIDIA/nvidia-azure-samples
   cd nvidia-azure-samples

2. Browse the Samples section below and pick the one that matches your use case.

3. Follow the README inside that sample's directory for setup and run instructions.

Samples

Samples are organized by use case:

• agentic/ - Agentic AI samples and reference workflows, including agentic/aks-samples/vss-on-aks/ for NVIDIA Video Search and Summarization (VSS) on Azure Kubernetes Service (AKS).
• inference/ - Model inference samples.
• training/ - Model training samples.
• data-processing/ - Data processing samples.
• physical-ai/ - Physical AI samples.
• industry-solutions/ - Industry-focused solution samples.

Usage

Each sample directory contains its own README with detailed deployment and usage instructions. In general:

1. Provision the required Azure infrastructure (cluster, compute, networking, storage, and AI services) as described in the sample.
2. Deploy the NVIDIA software components.
3. Run the included workloads or applications.
4. Clean up the resources when you are done to avoid unnecessary charges.

Known Issues

None at this time.

Contributions

Contributions are welcome. Developers can contribute by opening a merge request or pull request and agreeing to the terms in CONTRIBUTING.md.

Support

For questions or issues:

• Open an issue in this project for bug reports or feature requests
• Refer to each sample's README for sample-specific guidance

To report a security vulnerability, follow the process in SECURITY.md.

License

See LICENSE. This project is licensed under the Apache License 2.0.