All Posts
Onedata at EGI 2026: Unifying Distributed Data for Scientific and AI Pipelines EGI 2026
21–25 Sept.
Ghent
© Bart Laridon / CC BY-SA 4.0

Onedata at EGI 2026: Unifying Distributed Data for Scientific and AI Pipelines

At EGI 2026 in Ghent, Onedata showcased how a unified federated data space can bring together distributed research data and support scientific and AI workflows.

The annual EGI Conference brings together research and digital infrastructure providers, data scientists, policy-makers and innovators to advance data-intensive science. EGI 2026, organised by the EGI Foundation with the EGI Federation, took place on 21–25 September 2026 at the historic De Bijloke venue in Ghent.

Onedata was featured in two sessions demonstrating different aspects of federated data management — from practical interaction with research data services to integrating distributed data into scientific and AI workflows.

DataHub/Onedata for research data: integrate, compute, annotate, publish

The first session took place on Tuesday, 22 September, as part of the EGI Service Fair – EGI Services in Action.

In his 15-minute presentation, Łukasz Opioła introduced how Onedata addresses a fundamental challenge of modern research infrastructures: when data and computing resources are distributed across locations and organisations, accessing and working with that data should not become unnecessarily complicated.

The session also gave participants an opportunity to engage directly with the service owner, ask questions and discuss practical experiences, current challenges and future developments of the platform.

Tuesday, 22 September 2026, 16:00–16:15 CEST
EGI Booth Foyer, De Bijloke
Forum: Service Fair – EGI Services in Action
Speaker: Łukasz Opioła

Event details

Unifying Distributed Data for Portable Scientific and AI Pipelines with Onedata

On Wednesday, 23 September, Onedata was featured in a 30-minute demonstration titled “Unifying Distributed Data for Portable Scientific and AI Pipelines with Onedata.”

The session showed how Onedata provides a common data layer for scientific and AI workflows operating across distributed infrastructure. The presentation was based on research by Łukasz Opioła, Łukasz Dutka and Michał Orzechowski.

Wednesday, 23 September 2026, 17:30–18:00 CEST
Anatomical Theatre, De Bijloke
Forum: Data Management and Integration – Demonstrations 3
Format: Demonstration, max. 30 minutes
Speaker: Łukasz Opioła
Authors: Łukasz Opioła, Łukasz Dutka, Michał Orzechowski

Event details

AI agents and the data layer

On Thursday, 24 September, Michał Orzechowski expanded on this work in a talk about portable scientific and AI pipelines and the agents that run them. He showed how agents can access data and metadata through an Onedata mount, while MCP tools provide information about and control over the wider federation. Examples covered checking data distribution across providers, managing replication requirements, and verifying changes against the federation’s state. The talk also addressed scoped agent identities, human approval for consequential actions, and lessons from EOSC workflows.

Thursday, 24 September 2026, 15:27 CEST
Kraakhuis, De Bijloke
Session: Data Management and Distributed Research Environments (Part 2)
Format: Talk, 17 minutes + 3-minute Q&A
Speaker: Michał Orzechowski
Authors: Łukasz Opioła, Łukasz Dutka, Michał Orzechowski

Event details