Where science, AI, and infrastructure come together.

Areas of focus

We begin with the scientific question, then shape the approach and build the AI, software, and computing needed to answer it.

• AI-driven solutions for science and engineering. Foundation and domain models, AI agents, and surrogate models, trained and run at HPC scale.
• Cyberinfrastructure and distributed systems. Gateways, workflows, data movement, and identity, connecting laptops, clouds, and national supercomputers.
• Interpretable AI for scientific discovery. Methods that show what a scientific model has learned, from protein structure prediction to molecular simulation.

Projects

What we are building

• Nexus A very large AI supercomputing resource for open science, built and operated by Georgia Tech and the University of Illinois Urbana-Champaign. It gives researchers an integrated environment for moving from interactive development to large-scale workloads. nexus.gatech.eduDocsPortal
• CyberShuttle Connects researchers, applications, data, and workflows across local, cloud, and HPC resources, using familiar tools. cybershuttle.orgAI4Science Foundry
• AI2Science A research program on what foundation models actually learn about science, and how that structure can be tested and turned into scientific claims. ai2science.github.io
• Custos Open-source security for science gateways: federated login through CILogon, secure credential management, groups and sharing, and audit logs. Apache Custos
• VizFold Makes AlphaFold and similar protein-folding models interpretable, with interactive exploration in the browser. Funded under an NSF NAIRR EAGER award. vizfold.github.io
• SciGaP Hosted services for science gateways, including user management, workflow execution, and experiment archiving. Built on Apache Airavata. scigap.org
• SMILES Dense lattices of fluorescent dyes that avoid self-quenching, giving programmable, very bright emission. Project site
• AMOS Gateway Free access to state-of-the-art software for atomic, molecular, and optical science, through a coordinated international platform. amos-gateway.org
• Plasma VLab Virtual laboratory resources for plasma research, with computing access to the Gkeyll code. Project site

All projects

Software

Scientific software we maintain

• Apache Airavata — Open-source framework for managing and executing computational workflows and scientific applications on distributed HPC and cloud resources.
• Apache Custos — Open-source identity and access management for science gateways and research apps (OAuth2/OIDC, groups/roles, fine-grained authorization).
• Airavata MFT — High-performance, secure, multi-protocol data movement service for HPC, cloud, and gateways.
• Airavata Data Catalog — Metadata management and data discovery service that organizes, searches, and secures research datasets; integrates with common storage backends.

More on our software

Collaboration

How we work with science teams

A collaboration starts with the scientific problem, not the technology.

We partner with research teams at Georgia Tech and beyond to shape the approach, build the methods and software, and run them at the scale the science demands.

People

Who we are

Alumni: Tyler Hayes, AI Research Scientist, Arsenale Bioyards.

Publications

Selected publications

• Federated HPC Account Provisioning Using CILogon, COmanage, and PAM-Based OAuth2 Device Flow. Jayathilake, Jayawardana, Wannipurage, and Marru. PEARC '26, Article 67, 2026. doi:10.1145/3785462.3815896
• A Composable and Modular Framework for Protein Structure Prediction on HPC. Vennamreddy, Virani, Yin, Ramasubramanian, Ramasamy, Krishnan, and Marru. PEARC '26, 2026. doi:10.1145/3785462.3815886
• Δ-ML ensembles for selecting quantum chemistry methods to compute intermolecular interactions. Wallace, Sherrill, and Krishnan, 2025. arXiv:2511.17753
• Quantifying the role of OpenFold components in protein structure prediction. Hayes and Krishnan, 2025. arXiv:2511.14781
• Cybershuttle: An end-to-end cyberinfrastructure continuum to accelerate discovery in science and engineering. Marru et al. PEARC, 2023. doi:10.1145/3569951.3593602

All publications

Partners

Funding, industry, and research partners

Funding and support

Industry

Research institutions

More on our partners

Contact

Have a scientific challenge ?

Tell us what you are trying to solve. We would like to hear about it, and we are ready to collaborate towards scientific challenges. You do not need a finished plan or a funded project to get in touch.