Build Enterprise-Grade AI with NVIDIA AI Software

Following the introduction of ChatGPT, enterprises around the globe are realizing the benefits and capabilities of AI, and are racing to adopt it into their workflows.

As this adoption accelerates, it becomes imperative for enterprises not only to keep pace with the rapid advancements in AI, but also address related challenges such as optimization, scalability, and security.

The enterprise AI development journey typically begins with the data ETL (extract, transform, load) phase, during which data is prepared for training. This is followed by training the AI models. Once the models are trained, the next steps involve deployment and running inference. Enterprises need to use optimized and secure software for each of these stages to build production-ready AI applications.

Accelerate AI development with the NGC catalog

NVIDIA NGC catalog enables enterprises to accelerate their AI development by providing a suite of GPU-optimized software and SDKs in the form of containers.

Some of the popular containers in the catalog include RAPIDS for the data ETL phase, TensorFlow and PyTorch for the model development phase, and NVIDIA TensorRT and NVIDIA Triton Inference Server for the model deployment phase.

Figure 1 shows the different containers available for various use cases, such as NLP, object detection, recommendation, and more. Explore the NGC catalog to find the right software for your use case.

Screenshot of the available containers for various use cases on the NGC catalog.

Figure 1. The NGC catalog offers containers for a wide variety of use cases

Benefits of NGC containers

All the containers in the catalog are optimized to run on the latest NVIDIA accelerated computing stack and are updated regularly to provide enhanced performance on the same hardware. The containers are tested to ensure compatibility and performance across various NVIDIA GPUs.

Enterprises can deploy these containers anywhere, including on-premises and both multi-cloud and hybrid-cloud environments. The containers can be deployed with various container orchestration systems, including the widely used open-source platform Kubernetes. Additionally, the catalog offers a range of Helm charts to facilitate the deployment and management of containers on Kubernetes clusters.

Security and trust are ‌also an integral part of the containers and models available in the catalog. All ‌containers are scanned for CVEs and are assigned a security rating so that enterprises can be confident about the software they are downloading. Figure 2 shows the different security ratings for the containers.

Screenshot of the different security ratings available for the containers in the NGC catalog.

Figure 2. Different security ratings of the containers in the NGC catalog

Containers in general also provide several other benefits, such as:

• Encapsulation of dependencies: Containers encapsulate all the dependencies, such as libraries and other software components, to ensure that the application can run reliably in any environment without the need for additional configuration.
• Reproducibility: Containers contain everything needed to run an application, so they will behave the same way regardless of where they are deployed. This reduces the chances of issues related to environment-specific configurations.
• Time and resource savings: By simplifying the deployment process and ensuring consistency across environments, containers save considerable time and resources that would otherwise be spent on manual setup and troubleshooting.

Using the catalog software, enterprise developers can start building POCs and testing their solutions. When enterprises are ready to move into production, they need to ensure that the requirements for integrating AI into their business applications are met.

These requirements include enterprise-grade security, software stack stability, enterprise support, and manageability. Having a clear path to assist in transitioning from development to production is part of the success factor of time to value.

Seamless transition from development to production

The diverse range of software components and associated interdependencies make maintaining a secure, reliable AI software stack a massive undertaking, especially when AI needs to be deployed and integrated into enterprise applications. NGC catalog enables enterprise developers to seamlessly transition by offering both community-based software for development and enterprise-grade software branches for production.

For example, the catalog offers multiple containers for NVIDIA Triton Inference Server (Figure 3). NVIDIA Triton Inference Server Feature Branch is the latest version, released in a monthly cadence, to provide developers access to the latest features and performance optimizations.

Image showing the search results for NVIDIA Triton Inference Server query.

Figure 3. Different NGC containers are available for NVIDIA Triton Inference Server

NVIDIA Triton Inference Server Production Branch, available exclusively with NVIDIA AI Enterprise, is the production-grade version purposely built to provide stability and a secure environment for building mission-critical AI applications. The production branch releases every 6 months with a 9-month lifetime to ensure API stability. Within the 9-month lifecycle, NVIDIA continuously tracks critical vulnerabilities and releases CVE patches and bug fixes monthly without breaking the software stack.

With different optimized AI software options, enterprise developers can leverage the catalog to choose the proper software packages that support a given AI pipeline while maintaining security.

Uninterrupted AI excellence with NVIDIA AI Enterprise

Security, reliability, and manageability are critical for enterprise-grade AI. To address these challenges, NVIDIA has introduced NVIDIA AI Enterprise, an enterprise-grade software platform that accelerates the data science pipeline and streamlines the development and deployment of production-grade AI applications.

Built on open source and curated, optimized, and supported by NVIDIA, NVIDIA AI Enterprise offers multiple supported branches (production, feature, and long-term support). It also includes enterprise-grade security, stability, manageability, and support throughout your AI journey with NGC.

Summary

The NGC catalog and NVIDIA AI Enterprise offer enterprises the necessary tools to keep pace with the rapid advancements in AI while addressing related challenges such as optimization, scalability, and security.

Visit the NGC catalog to browse hundreds of pretrained models, containers, Helm charts, and Jupyter Notebooks. To get started with NVIDIA AI Enterprise, apply for a free 90-day trial.

By Nirmal Kumar Juluru and Phoebe Lee/MVDIA