For Instructors · Submission Helper

Request a NAIRR Pilot Start-Up Allocation

What the request form needs, plus ready-to-adapt text for an education / outreach project. Replace the highlighted parts with your details and you're most of the way there.

📝

Where to submit

Start your request at nairrpilot.org → Start-Up Project (it routes you to the submission portal). Sign in with the identity you registered with — often "ACCESS CI (XSEDE)". Approvals typically come in a few days.

Prefer plain text? Everything below is also in 055a-startup-request-template.md for easy copy-paste into the form.

What the request asks for

Item	What to provide
Project title	Short, descriptive (it's posted publicly). See template below.
Abstract	A public summary of the project. Template below.
Research description / justification	What you'll do and how it aligns with NAIRR's goals (education, broadening access). Template below.
PI info	Name, affiliation, institutional email (not Gmail).
Resource (pick ONE)	For this workshop: Indiana Jetstream2 — up to 128,000 SUs.
Funding info	List any grants directly supporting the work (or note none).
CV / advisor letter	PI CV; if you're a grad student, a support letter from your faculty advisor.

📏

Know the limits going in

Start-Up projects run 3 months, request one resource, and require a U.S. institution and institutional email. Results must be open / publishable. Jetstream2's Start-Up cap is 128,000 SUs — far more than a CPU teaching workshop needs (see the estimate below).

Ready-to-adapt text

Project title

Keep it concrete and public-friendly.

Hands-On Introduction to NAIRR & Jetstream2 for [Your Institution] Students and Educators

Abstract (public)

~100–150 words. This is posted on the NAIRR website, so write for a general audience.

This project delivers hands-on workshops that introduce students and faculty at [Your Institution] to the National AI Research Resource (NAIRR) and the Jetstream2 cloud computing system. Participants learn to launch cloud instances, run Jupyter notebooks, and compare how AI models perform across multiple inference frameworks — gaining practical experience with national AI cyberinfrastructure they would otherwise be unable to access. By lowering the barrier to entry and openly publishing all teaching materials, the project broadens participation in AI research, strengthens AI readiness among the next generation of researchers and educators, and shows faculty how straightforward it is to bring NAIRR resources into their own courses.

Research description / justification

Explain the activity, the audience, alignment with NAIRR's goals, and the outcomes.

Motivation. Many institutions and educators lack access to the GPU and cloud computing needed to teach and explore modern AI, and many faculty assume that obtaining national resources is too complex to attempt. This project removes that barrier by running practical, reproducible workshops on NAIRR's Jetstream2 resource. Activities. We will deliver [number] hands-on workshop sessions for approximately [20] participants each. Using openly published Jupyter notebooks, participants launch their own Jetstream2 instances and run guided exercises — including a comparison of local AI inference frameworks (Ollama, llama.cpp, and vLLM) running the same model — to understand cloud-based AI computing and the trade-offs between tools. Sessions cover both classroom and research use, and how to apply for NAIRR resources independently. Alignment with NAIRR goals. The project directly advances NAIRR's mission to broaden access to AI cyberinfrastructure and build the AI-capable workforce: it reaches learners and educators at [institution type, e.g., a regional public university serving underrepresented students], and it equips faculty to sustain AI education beyond this project. Outcomes. All notebooks, slides, and step-by-step guides are published openly on GitHub for reuse by any educator. We will report participation and lessons learned in the required final report.

Resource & SU justification

Pick Jetstream2 (IU). Show the math so the amount is obviously reasonable.

Requested resource: Indiana University Jetstream2. Each workshop runs ~[20] participants, each on a small CPU instance (m3.quad, 4 vCPUs) for ~[3] hours. At roughly 1 SU per vCPU-hour, that is about 4 SU/hour per instance, or ~240 SU per session for 20 participants. Across [number] planned sessions plus instructor preparation and testing, total usage is on the order of a few thousand SUs — well within the 128,000 SU Start-Up cap, with comfortable headroom for repeat sessions and occasional GPU experimentation.

✍️

Tips that help approval

• Lead with education / broadening access — that's squarely in NAIRR's mission.
• Emphasize that outputs are open (public GitHub repo, final report).
• Keep the resource ask modest and justified — a CPU teaching workshop needs very little.
• Have your CV ready to upload, and use your institutional email throughout.

➡️

After it's approved

You'll get an award email with your project ID and SUs. Then follow the allocation playbook to add your students and run the class.

🌱

Thinking ahead: your full proposal

The Start-Up is your on-ramp. Once you've run a session or two, the natural next step is a full NAIRR Pilot proposal (a ~3-page request, 12-month allocation) to sustain and grow the work. It helps to start sketching it early — what you'll do, who you'll reach, and the resources you'll need.

And you don't have to start from a blank page: if you set up VS Code + Claude (see the AI assistant guide), the same assistant that builds notebooks can help you draft the proposal — describe your project and let it draft sections you then refine.