Templates enable platform teams to build automation for cloud environments and Kubernetes, jumpstarting adoption of Large Language Models (LLMs), while supporting traditional AI applications leveraging SLURM, MLFlow and Kubeflow
Rafay’s GenAI templates empower platform teams to efficiently guide GenAI technology development and utilization, and include reference source code for a variety of use cases, pre-built cloud environment templates, and Kubernetes cluster blueprints pre-integrated with the GenAI ecosystem. Customers can easily experiment with services such as Amazon Bedrock, Microsoft Azure OpenAI and OpenAI’s ChatGPT. Support for high-performance, GPU-based computing environments is built into the templates. Traditional tools used by data scientists such as Simple Linux Utility for Resource Management (SLURM), Kubeflow and MLflow are also supported. Developers and data scientists attending Kubecon 2023 in Chicago this week can see live demonstrations at booth C31. All templates, reference designs and sample code are available in Rafay’s Git Repository and Rafay’s Documentation.
According to Gartner, by 2026, more than 80 percent of enterprises will have used generative AI APIs, models and/or deployed GenAI-enabled applications in production environments, up from less than 5 percent in 2023. In the dynamic landscape of modern technology, platform teams find themselves at the forefront of a transformative journey. They are tasked with a pivotal role – to not only initiate but also to set clear guidelines for the development and integration of GenAI technologies.
Rafay’s GenAI templates simplify the environment setup for deploying AI applications so platform teams can take the lead in AI development, and empower developers and data scientists to harness the full potential of AI. Developers gain a competitive advantage by being able to press a button and consume an enterprise-grade AI development sandbox in a controlled self-service manner, expediting the innovation process. Rafay’s native offering continues to deliver the control and efficiency platform teams need to maintain oversight while keeping costs in check.
“As platform teams lead the charge in enabling GenAI technologies and managing traditional AI and ML applications, Rafay’s GenAI focused templates expedite the development and time-to-market for all AI applications, ranging from chatbots to predictive analysis, delivering real-time benefits of GenAI to the business,” said Mohan Atreya, Rafay Systems SVP of Product and Solutions. “Platform teams can empower developers and data scientists to move fast with their GenAI experimentation and productization, while enforcing the necessary guardrails to ensure enterprise-grade governance and control. With Rafay, any enterprise can confidently start their GenAI journey today.”
Unlocking the Full Potential of AI with a Controlled Self-Service Approach
Rafay’s GenAI templates deliver autonomy to developers and data scientists, while streamlining the integration and resource management of AI infrastructure, such as cloud environments and Kubernetes clusters. Enterprise platform teams benefit from the following capabilities:
- Self-Service Experience: Developers and data scientists can deploy, view and manage their GenAI applications and infrastructure in isolation using self-service workflows via Rafay and Spotify Backstage.
- AI/ML Ecosystem Support: Rafay provides out of the box support for LLM providers including Amazon Bedrock, Azure OpenAI and OpenAI.
- AI Applications and Source Code: Several GenAI and AI workbench applications with source code such as a text summarization and a chatbot app using GenAI are included.
- Any Orchestration, Any Cloud: Pre-built templates for Amazon ECS, EKS/A, Microsoft AKS and Google GKE on public clouds as well as private data centers and edge locations help streamline AI resource management.
- Cluster and Workflow Standardization: Rafay’s Environment templates for Kubernetes blueprints allow platform teams to create a set of standard GenAI environments and make them available enterprise-wide.
- Secure RBAC: Each developer, data scientist, researcher, etc. can create and destroy environments (but not templates built by platform teams) and operate them in isolation, governed by RBAC.
- Integrated GPU and Kubernetes Metrics: Rafay automatically captures and aggregates both Kubernetes and GPU metrics at the controller in a multi-tenant time series database.
- Multitenancy for AI/ML Apps: It is incredibly common for enterprises to have different teams share clusters – perhaps with specific LLM resources – in an effort to save costs. Rafay’s multi-modal multi-tenancy capabilities can easily support multiple AI/ML teams on the same Kubernetes cluster.
- Chargeback & Showback: Rafay provides each isolated unit financial metrics including chargeback and showback for their AI applications across private and public clouds
- Support for Traditional AI Platforms: Rafay also supports traditional AI frameworks such as SLURM, KubeFlow and MLflow.
SOURCE: BusinessWire
