NVIDIA GTC – Outerbounds, a developer platform for ML/AI systems built by the creators of Metaflow that powers Netflix and hundreds of other sophisticated ML/AI organizations, announced that select customers can now access NVIDIA DGX Cloud, available through the NVIDIA ecosystem of leading cloud service providers. As a result, companies previously limited by the cost or availability of cloud-based accelerated computing have a new avenue to run their experiments and production workloads confidently at scale.
Innovation shouldn’t be limited by compute
As companies are increasingly adopting sophisticated data-intensive techniques to improve their business, including both traditional machine learning as well as new large language models and generative AI, their compute requirements grow rapidly. The new workloads demand new kinds of software stack and tooling, such as the platform provided by Outerbounds.
With the new integration, Outerbounds’ customers can now use serverless APIs to easily deploy and manage AI tasks on GPUs via NVIDIA DGX Cloud. This offers better security, scalability, and reliability for their workloads, both during prototype development and production runs of new ML and AI models.
Thanks to Metaflow, known for its human-centric design, ML and AI developers and data scientists can benefit from the new capability with just a few lines of code, empowering them to experiment faster and ship to production quicker. Meanwhile, the engineering teams can rest assured that production workloads won’t run into capacity issues, which have been a common concern over the past years.
NVIDIA DGX Cloud
The compute resources are available through NVIDIA DGX Cloud, which provides access to accelerated computing infrastructure, tools, and software that streamline developer workflow and scale productivity. DGX Cloud enables an end-to-end AI development experience, from pretraining to fine-tuning to inference, powered by a serverless API to deploy and manage AI workloads, with enterprise-grade security, scale, and reliability.
Build and ship ML/AI systems, not just models
Building real-world, production-quality ML and AI systems takes more than access to GPUs. In order to build real-world ML and AI systems, companies need to be able to:
- Access a deep pool of compute resources.
- Read and write data quickly and securely from their data warehouses.
- Orchestrate even advanced workloads in a highly available manner.
- Integrate to their surrounding systems seamlessly.
- Keep track and observe all experiments and production workloads.
- Provide a productivity-boosting development environment for their teams.
Accomplishing all of the above in a secure, compliant, and cost-effective manner, running securely in their cloud account. By accessing Outerbounds, powered by DGX Cloud, companies can realize all of the above benefits without having to spend engineering resources on infrastructure.
SOURCE: PRNewsWire