JFrog & NVIDIA team up to enhance secure AI deployments

JFrog has integrated NVIDIA NIM microservices into its Software Supply Chain Platform, enabling secure and efficient deployment of AI models such as Meta's Llama 3 and Mistral AI.

This integration is designed to streamline AI deployments by aligning DevSecOps workflows with enhanced security and compliance measures. Enterprises are now offered a secure pathway to adopt AI, optimising performance for scalable applications and addressing industry demands for higher security standards.

Gal Marder, Chief Strategy Officer at JFrog, said, "The demand for secure and efficient AI implementations continues to rise, with many businesses aiming to expand their AI strategies in 2025. However, AI deployments often struggle to reach production due to significant security challenges. AI-powered applications are inherently complex to secure, deploy, and manage, and concerns around the security of open-source AI models and platforms continue to grow."

"We're excited to collaborate with NVIDIA to deliver an easy-to-deploy, end-to-end solution that enables companies to accelerate the delivery of their AI/ML models with enterprise-grade security, compliance, and provenance."

As AI becomes increasingly integral to software applications, data scientists and ML engineers face challenges scaling enterprise ML model deployments. Complexities in integrating AI workflows with existing development processes, alongside issues such as fragmented asset management and compliance vulnerabilities, can lead to extended deployment cycles. IDC projects that by 2028, 65% of organisations will employ DevOps tools combining multiple Ops capabilities to optimise software delivery processes.

Jim Mercer from IDC commented, "The rise of open-source MLOps platforms has made AI more accessible to developers of all skill levels, allowing them to quickly build amazing AI applications, but this process needs to be done securely and in compliance with today's quickly evolving government regulations."

"As enterprises scale their generative AI deployments, having a central repository of pre-approved, fully compliant, performance-optimised models developers can choose from and quickly deploy while maintaining high levels of visibility, traceability, and control through the use of existing DevSecOps workflows is compelling."

The integration between JFrog and NVIDIA NIM aims to facilitate the seamless deployment and management of foundational LLMs, maintaining security and governance throughout the software supply chain. JFrog Artifactory, the key to the JFrog Platform, offers a unified solution for managing software artefacts, including binaries and ML Models, during the software development lifecycle.

The platform enables developers to access NVIDIA NGC, a hub for GPU-optimised deep learning models. It provides a unified source for software models and tools while ensuring visibility and control through DevSecOps best practices.

The JFrog Platform's update brings several advantages to AI developers and DevSecOps teams, such as unified ML and DevOps workflows that allow data scientists to use familiar processes. The platform provides automated compliance checks, audit trails, and model governance via JFrog Curation, eliminating the need for separate ML tools.

It also enhances security through continuous scanning for vulnerabilities across containers and dependencies, offering smart threat detection and proactive protection against threats.

The integration claims to boost model performance and scalability by utilising NVIDIA's accelerated computing infrastructure, accommodating low-latency deployments, and facilitating the bundling of ML models with dependencies.

Pat Lee, Vice President of Enterprise Strategic Partnerships at NVIDIA, remarked, "Performance and security are crucial for successful enterprise AI deployments. With NVIDIA NIM integrated directly into the JFrog Platform, developers can accelerate AI adoption with a unified, end-to-end solution for building, deploying, and managing production AI agents at scale."