Early Access to Union Serverless, Empowering Scalable AI Workflows

DateJun 10, 2024

Union.ai has unveiled early access to Union Serverless, a new cloud-based AI orchestration platform designed to simplify running production-grade AI workflows at scale without requiring infrastructure setup. Union, known for the open-source AI Orchestrator Flyte, aims to unify work streams across machine learning, data, and operations teams.

This service addresses challenges in model selection, fine-tuning, and infrastructure management by offering robust, reproducible workflows that operate seamlessly both locally and in the cloud.

Union Serverless allows developers to run various AI workloads such as fine-tuning off-the-shelf models, batch inference, and model training workflows effortlessly. It provides tools for validating datasets, versioning training code, building container images, and allocating appropriate resources, ensuring a frictionless experience.

Ketan Umare, CEO of Union, stated, “Union Serverless is a critical step towards democratizing AI infrastructure, making it accessible to organizations regardless of their existing capabilities. Our goal is to empower teams to become self-sufficient AI powerhouses.”

Union Serverless also supports a ‘bring your own cloud’ (BYOC) model, managing resources within a customer’s VPC. This flexibility has already attracted enterprises and startups, including Woven by Toyota, Flawless, and ArteraAI, which leverage Union’s platform for various applications. Union Serverless provides access to GPUs and other resources by the minute, with the option to migrate to Union BYOC for full control over data locality and cloud resources.

Executing a Machine Learning Pipeline

This Union Serverless platform is designed to enhance the productivity of AI developers by allowing them to focus on model development without worrying about deploying and managing code at scale. The lightweight Python library provided by Union Serverless enables application developers to run code on specialized GPUs, integrate proprietary datasets, and incorporate third-party APIs like ChatGPT, fostering rapid development of new AI applications.

Grantham Taylor, ex-Principal Data Scientist at Capital One, praised the platform’s ease of use, highlighting how he quickly connected to his Serverless namespace and executed an ML pipeline without dealing with complex infrastructure management.

Union Serverless is poised to revolutionize the way AI workflows are managed, offering a streamlined, scalable solution that empowers developers and organizations to harness the full potential of AI technology with minimal overhead. This innovation aligns with Union’s long-term vision of creating a unified infrastructure that leverages both customer cloud resources and Union’s pooled resources, all through an intuitive user interface.

Leave a Reply