Terrell Russell, Ph.D.

Executive Director, iRODS Consortium

November 16-21, 2025

Supercomputing 2025

St. Louis, MO

Abbreviated History

We're 30 this year!

• From science project to hardened code
• From philosophy to platform to product

Open Source

• Multiple releases
• Feature consolidation
• New members
• Market clarity
• Partnerships

AI requires good data management

AI and other analytics tooling do not provide good value if the data being fed into the systems are not good.

This is true - before, during, and after the analytics are performed.

Ensure good data through good data management practice:

• early automated ingest, clean sourcing
• strong provenance / chain of custody
• quality assessment, evaluation, cleaning, preparation
• access control and policy enforcement
• data staging

Preventing "Garbage In"

These best practices afford opportunities to:

• meet compliance requirements
• ensure integrity of files and data products
• reuse data for training the next models
• connect agentic assistants to any pipeline stage
• capture / associate metadata for better discoverability

All these can be automated with iRODS.

Preventing "Garbage In" through iRODS

Using iRODS as your data store provides

• programmability
• flexibility
• insurance against technology churn
• a way to avoid vendor lock-in

And it's open source.

The iRODS Consortium provides membership, service, support, consulting, design, engineering, market analysis, and partnership for new products and platforms.

Areas of active collaboration

• Metadata templates and validation
• Metrics, visibility, forecasting
• AI integrations
• Policy based safety / compliance
• Managed services

Thank you!

Booth #4424