AI4EOSC: Federated Cloud Platform for AI in Scientific Research
The article introduces AI4EOSC, an open-source federated platform aimed at implementing the complete AI/ML lifecycle in the European Open Science Cloud (EOSC) framework. It fills the void between conventional MLOps tools and the specific needs of contemporary Open Science, especially regarding FAIR principles. This platform combines a modular design with an AI development environment, serverless AI-as-a-Service, and federated orchestration to harmonize diverse computing and storage assets. Additionally, AI4EOSC adopts a 'FAIR-by-design' strategy that mandates metadata standardization through MLDC.
Key facts
- AI4EOSC is a federated open-source platform for AI/ML lifecycle in EOSC.
- It addresses the gap between MLOps tools and Open Science FAIR principles.
- The platform integrates modular architecture with AI development, serverless AI-as-a-Service, and federated orchestration.
- It unifies heterogeneous compute and storage from distributed e-Infrastructures.
- AI4EOSC enforces metadata standardization via MLDC.
- The paper is published on arXiv with ID 2512.16455.
- The platform is designed for the European Open Science Cloud ecosystem.
- It operationalizes the full AI/ML lifecycle.
Entities
Institutions
- European Open Science Cloud
- arXiv