NSF-supported · Early development

One interactive laboratory across the nation's research systems.

A unified environment that connects researchers, educators, and students to interactive computing and data across five NSF-supported NAIRR and ACCESS sites and two cloud providers.

National reach

Five HPC/AI sites

NSF-supported NAIRR and ACCESS computing sites, brought together in one interactive environment.

Elastic cloud

Two cloud providers

Two cloud providers will support elastic and hybrid workflows when cloud capacity or services are the right fit. Providers will be announced once integration planning is complete.

One data layer

Shared data across resources

A shared data layer so researchers can work with datasets where they already reside.

FAIR outputs

FAIR apps & publications

A consistent application catalog and FAIR-aligned publishing.

What iDLab enables

One interactive front door to research systems nationwide.

Research increasingly depends on interactive, data-intensive, and AI-enabled work to enable rapid iteration across large datasets. iDLab brings these powerful NSF-supported systems together in a single, browser-based environment, adding a consistent interactive layer on top of the resources researchers already rely on, making them easier for more people to access.

What it is

A single, browser-based environment for interactive computing, a consistent application catalog, and shared data access across partner ACCESS and NAIRR systems and cloud providers.

What it will provide

Exploration where the data already lives, single sign-on across facilities, collaborative workspaces, and FAIR publishing of results.

What it is not

Not a long-term archive. iDLab builds on participating NSF cyberinfrastructure by adding a shared, browser-based layer for applications, data access, and cross-site workflows.

iDLab advances OneSciencePlace from a pilot toward a national-scale service. This is the start of that work: integration, testing, and a structured early-user phase come first, followed by continuous operation.

What the platform offers

Interactive work, from one place.

A web interface designed so that newcomers and power users alike can access advanced systems without site-specific configuration.

Interactive sessions

Launch GPU- and CPU-backed interactive applications in the browser for analysis, visualization, modeling, and AI/ML development.

Consistent app catalog

A unified set of interactive tools, including Jupyter, RStudio, and visualization and analytics apps, is consistently deployed across sites.

Global Data Namespace

A consistent way to name, find, and access data across connected systems. Data can remain at an ACCESS or NAIRR site or in the cloud, while iDLab provides a common access layer that reduces manual staging and unnecessary copies across workflows.

Collaborative workspaces

Shared project spaces with role-based access, enabling teams to organize data, applications, and provenance for reproducible science.

FAIR publishing

Publish datasets, workflows, and apps with FAIR-aligned metadata and persistent identifiers for sharing and reuse, not for long-term archival.

Federated identity

Single sign-on via CILogon and Globus Auth, mapped to site accounts to keep access consistent and secure across facilities.

Where iDLab runs

Five HPC/AI sites, two clouds, one environment.

iDLab integrates with each site's local policies and adds project-dedicated interactive resources. Each site keeps full operational control of its systems.

Lead site

UCLA OARC

Office of Advanced Research Computing · University of California, Los Angeles

Project leadership, platform development, integration, and coordination.

NCSA

National Center for Supercomputing Applications

Urbana-Champaign, IL

Systems · Delta / DeltaAI

PSC

Pittsburgh Supercomputing Center

Pittsburgh, PA

Systems · Bridges-2

Purdue

Purdue University

West Lafayette, IN

Systems · Anvil / Anvil AI

SDSC

San Diego Supercomputer Center

UC San Diego, CA

Systems · Expanse / Expanse AI

TACC

Texas Advanced Computing Center

UT Austin, TX

Systems · Stampede3

Cloud

Two cloud providers

To be selected

Mode · Elastic HPC-cloud

The research data lifecycle

From ingestion to publication.

iDLab is built around the work researchers actually do: bringing data in, organizing it, exploring it, sharing it, and publishing results for reuse.

  1. 1

    Ingest

    Upload, generate from simulations, or move data in via Globus and the shared layer.

  2. 2

    Organize

    Shared workspaces hold data, apps, and job provenance.

  3. 3

    Explore

    Interactive sessions support iterative, human-in-the-loop work.

  4. 4

    Share

    Role-based access enables secure team science across institutions.

  5. 5

    Publish

    FAIR-aligned outputs with persistent identifiers.

iDLab supports short-term publication for sharing and reuse, and it is not a long-term archive. At the end of the project, data can be moved to a domain or another repository.

Powering discovery & learning

Built for data-intensive, interactive research.

iDLab supports fields where interactive analysis of site bound, distributed datasets is essential. It gives researchers a faster, more hands-on way to work with the systems that store their data. A few representative areas:

Engineering

Natural hazards engineering

Earthquake simulation and seismic hazard modeling, with cross-site access to high-fidelity rupture and ground-shaking data.

Life sciences

Spatial biology

Tissue and cellular imaging at unprecedented scale, with interactive tools to evaluate, annotate, and prepare data for AI.

Neuroscience

Brain imaging & signals

AI-driven analysis of imaging and signal data, with computation placed alongside the datasets it depends on.

Geosciences

Geospatial analytics

Mapping and spatial AI for extreme-event response and resilience, on dedicated GPUs and a unified data layer.

Education

University courses

Hands-on computational learning for STEM students in ready-to-use, browser-based environments beyond notebooks alone.

For the research and education community

These are just a few examples. iDLab is built for the broad research and education community, not only the fields shown above.

From cryo-EM and computational physics to digital agriculture, nuclear engineering, and university classrooms wherever interactive, data-first work happens, iDLab could help support it.

Get involved

Help shape iDLab from the start.

We're inviting researchers, educators, and resource partners to follow the project and take part in early testing as iDLab comes online. If your work depends on interactive analysis and distributed data, we'd like to hear from you.