Overview:
The National Research Platform (NRP) is a free, shared, Kubernetes-based research computing resource available to SLU. Funded by the National Science Foundation (NSF) and community-governed, NRP connects more than 5,000 researchers at over 70 institutions worldwide. It enables researchers to pursue data-intensive projects with access to high-performance GPUs, CPUs, storage, and the tools to run their most demanding workflows.
This article provides a plain-language overview of NRP and its resources. For setup and access instructions, including kubectl, JupyterHub, and storage, visit the NRP documentation.
Who Runs It & How It's Governed:
- NSF-funded: NRP operates through a series of NSF awards, including CNS-1730158, ACI-1540112, ACI-1541349, OAC-1826967, OAC-2112167, CNS-2100237, and CNS-2120019. Participating institutions do not pay for compute time.
- Community-governed: NRP describes itself as “shared research computing, run by the people who use it.” Member institutions make decisions about the platform rather than a single company or agency.
- Organized in three tiers: Organizations, such as universities and research consortia, contain Labs, typically led by a faculty member or principal researcher. Labs contain Projects, each of which maps to a Kubernetes namespace; an isolated workspace for a research group.
- Group-level workspace management: Each namespace has an administrator who works in the research group and helps manage its resources and workloads. This keeps decisions close to the people using the workspace and gives each Lab direct control over its own resources and workloads.
- Federated sign-in: Use your existing university account to log in through CILogon. First, ask your research supervisor to add you to an existing namespace or request a new one affiliated with Saint Louis University by following the NRP access instructions.
Compute & Hardware:
NRP’s resource pool includes general-purpose processors, accelerators, storage, and networking options. Availability varies across the platform, so check the NRP resources page for current hardware details.
- CPUs: Nodes range from 16 to 384 CPU cores, supporting both interactive work and large parallel jobs.
- GPUs: Multiple GPU models are available, with up to 16 GPUs on a single node. Options vary by model and location, with as many as 16 GPUs on a single node for the most demanding jobs.
- Specialized hardware: FPGAs and other accelerators are available for networking research and more advanced hardware work, alongside the general-purpose CPU/GPU pool.
- Storage: Petabyte-scale storage capacity, spread across several different backends suited to different needs (fast block storage, shared filesystems, S3-compatible object storage, and more). See the NRP storage documentation for details.
- Networking: Sites are connected by a high-speed backbone, with automatic TLS-secured routing (your-project-name.nrp-nautilus.io) for anything a researcher exposes as a web service.
Ways to Use the Cluster:
There are multiple ways SLU researchers can interact with the Nautilus cluster, ranging from browser-based access requiring no setup at all to full command-line control. The Nautilus user guide outlines all the available options, including JupyterHub notebooks, a simple GUI integration for VS Code, or the kubectl command line tool for full control. Use these quick start guides to get going:
A complete list of all hosted software & services can be found here.
AI & LLM Hosting:
NRP offers free access for researchers and educators to a rotating catalog of frontier open-weight models, with no per-token billing. Current models include Qwen3, Kimi, Gemma, GLM-5, DeepSeek, MiniMax, and OpenAI's GPT-OSS. Some models are labeled "evaluating." For the latest model status, see NRP's Lifecycle & changelog.
Access is available through a hosted chat interface (Open WebUI), ready-made coding-CLI configs (including Claude Code), and an OpenAI-compatible API. The NRP documentation for Managed LLMs includes setup details and examples of uses such as document understanding, agentic coding for classrooms and research groups, batch processing over large document collections, and troubleshooting cluster workloads.
Education & Community Programs:
Instructors at U.S. nonprofit institutions, including community colleges, can set up their whole class with its own namespace and GPU access, no hardware to buy or maintain. Students can train models at the same time instead of taking turns on a single GPU. To get started, contact NRP with your course details, expected student count, and start date before the first class session.
Beyond the classroom, NRP offers community office hours, recorded training, and a community workshop series. NRP is also named in the NSF AI Infrastructure Hubs solicitation as a way for Hubs to integrate their resources, and it supports LightScope, a distributed network-telescope project for cybersecurity research.
Support, Documentation & Disclaimer:
SLU ITS Research Technologies can point you to the right NRP resource and help with getting started, including requesting a namespace. Nautilus itself is owned, operated, and supported by the National Research Platform, not Saint Louis University. For questions about specific services, hardware availability, policies, or problems with your jobs, contact NRP Support directly.