Synopsis

Enterprise AI Engineers are rapidly shifting from model experimentation to deploying full-stack AI systems that are secure, reproducible, and cloud-native. This session explores how open source tools—RamaLama, Podman, and Llama Stack—empower developers to streamline the development and deployment of AI applications featuring Retrieval-Augmented Generation (RAG) and containerized inference.

We’ll demonstrate how RamaLama provides a composable and secure runtime environment for AI models that runs in isolation by default, making “boring” production deployments possible without bespoke infrastructure or MLOps overhead. Using Podman and OCI containers, developers can securely distribute and execute models across environments. We’ll also show how Llama Stackunifies APIs for inference, agents, telemetry, and RAG pipelines, accelerating the development of agentic AI applications.

What you will learn: (a few bullet points):

• Run models with fine-grained security and sandboxing using RamaLama
• Seamlessly integrate RAG workflows with data pipelines
• Package and deploy AI workloads using OCI-compliant containers via Podman
• Leverage Llama Stack to prototype and deploy secure, modular AI systems across hybrid cloud environments

By combining container-native isolation with unified development interfaces, this talk offers a practical blueprint for building scalable, trustworthy AI systems—without locking developers into a single platform or vendor.

Note:
Red Hat is committed to the health and safety of our customers, partners, and employees. All attendees will be expected to follow Red Hat and local guidelines regarding COVID-19. Please do not attend if you are feeling ill or have any symptoms of a viral infection. Please note that all protocols are subject to change.