FLUID-AI
FLUID-AI
FLUID-AI addresses the lack of interoperability among data, AI, machine-learning models, and digital solutions within the European Open Science Cloud (EOSC). It introduces the concept of Data and Models Liquidity, extending the FAIR principles to meet the specific needs of AI-ready data and models. In addition to being findable, accessible, interoperable and reusable, AI-ready assets must be structured, annotated, technically compatible and optimised for integration into AI and machine-learning workflows.
FLUID-AI tackles three key gaps in the current EOSC ecosystem. First, it will establish a collaborative Competence Centre that provides coordinated support, training, and resources for researchers, research-infrastructure operators, and service developers. Second, it will develop semantic and technical interoperability solutions to enable the discovery, combination and reuse of data and AI models across platforms, research infrastructures and scientific disciplines. Third, it will deliver accessible platforms, tools and automated workflows that reduce technical complexity and enable researchers to focus on scientific discovery.
The project will validate its approach through eight use cases involving representative research infrastructures and data communities. They will demonstrate the scalability, reproducibility and practical relevance of the project’s solutions, contributing to the development of a dynamic, trustworthy and AI-ready EOSC ecosystem.
EGI Foundation role in the project:
EGI contributes to the project by building on its expertise in federated e-infrastructures, EOSC services, cloud computing, AI platforms and the integration of distributed research resources.
EGI will contribute in particular to:
- Supporting the integration of AI and machine-learning services into the EOSC ecosystem and the coordination with related EOSC initiatives and projects, like EOSC Beyond and EOSC Data Commons;
- Contributing to the development of interoperable platforms and services for the AI and machine-learning lifecycle, including data preparation, model development, training, deployment, inference and monitoring.
- Supporting the integration of distributed and heterogeneous computing resources through federated infrastructure services.
- Contributing to the development of service-mesh capabilities for AI-model inference across the cloud–compute continuum.
- Helping connect repositories, data services and computing resources so that AI-ready datasets can be accessed and processed efficiently.
- Supporting the adoption of open standards, APIs, interoperable protocols and reusable reference implementations.
- Contributing to validating project tools and services through real-world research-infrastructure use cases.
Expected results:
FLUID-AI will deliver:
- A framework for Data and Models Liquidity, extending FAIR principles to support AI-ready and machine-actionable data and models.
- Interoperability tools and metadata solutions enabling the discovery, integration and reuse of data and AI models across research infrastructures and disciplines.
- An accessible EOSC AI/ML platform with open-source tools, automated workflows and MLOps support across the AI lifecycle.
- Federated services for AI-model development, training, deployment and inference across distributed computing resources.
- A blueprint and guidelines for trustworthy AI-ready repositories.
- A Competence Centre offering training, support and learning resources for researchers, data stewards, infrastructure operators and service developers.
- Validation of the solutions through eight real-world use cases involving European research infrastructures and data communities