Technology exchange
Tutorials

Building AI-Ready Campus/Cloud Infrastructure: MLOps for AI-driven Scientific Workflows on AWS | 8–11:30 a.m.

This tutorial provides a strategic and technical roadmap for building “AI-ready” campus and cloud infrastructure capable of supporting the next generation of data-intensive AI-based scientific workflows. As higher education institutions scale their investments in GPU-backed cloud resources, IT leadership faces the critical challenge of ensuring these assets lead to reproducible and efficient scientific outcomes. To address this, IT teams must move beyond basic virtual machine provisioning and adopt DevOps for scientific workflows.

Attendees will learn how to bridge the operational gap between raw cloud instances and managed research pipelines using the Mizzou Cloud DevOps (MCD) online learning platform. We demonstrate a practical approach to operationalizing research by implementing MLflow on AWS as a centralized service for experiment tracking and model management. Through a hands-on lab, we show how system and network engineers can transform complex, manual research tasks, specifically Active Learning, into structured, repeatable cloud-native pipelines. By the end of this tutorial, participants will understand how to build the foundational infrastructure necessary to maximize the impact of institutional AI investments while ensuring resource stewardship and workflow automation. Please check the description in the online program for a full description of all elements covered in this tutorial.

Hands-on Tutorial on SmartNICs / Data Processing Units (DPUs) | 8–11:30 a.m.

This tutorial, organized by the University of South Carolina and Internet2, will provide an effective hands-on training on SmartNICs. The tutorial will cover offloading core infrastructure workloads to the domain-specific accelerators of the SmartNICs. The content is designed for CI engineers, network administrators, and research computing professionals. Participants will gain hands-on experience in configuring and managing a wide range of DPU-based workloads, including packet forwarding, firewalling, encryption, decompression, and storage acceleration. The labs leverage the NVIDIA BlueField DPU and are accessible through a web-based environment.

MetrANOVA User Hands-on Tutorial | 8–11:30 a.m.

This tutorial introduces MetrANOVA, a new software stack designed to monitor flow, SNMP, and streaming telemetry data using a robust suite of free and open-source tools. Developed and deployed by a consortium of Research and Education networks, MetrANOVA is now being made available to the wider R&E community.

In this hands-on instructional lab, the development team will guide attendees through the core architecture of the stack, including its use of ClickHouse for high-performance data storage and Grafana for visualization, all running within Docker and Kubernetes environments.

This tutorial is designed for a General Audience; no prior experience with MetrANOVA or underlying components is required. Participants will move from a foundational project introduction to direct interaction with the software, gaining the practical skills needed to deploy and manage the stack within their own institutions.

Practical AI for Technical Professionals: A Hands-On Tutorial | 8–11:30 a.m.

The hype around AI code generation is real — and largely overstated. But something less glamorous and far more immediately useful is happening for engineers who’ve figured out how to use AI as a cognitive prosthetic: offloading brain static, breaking writer’s block, interrogating their own notes, and building a persistent second brain that knows their job.

This tutorial is a practitioner-to-practitioner walkthrough of that system, built live in an R&E context.

Research & Data Science Strategy for IT Leaders and Engineers | 8–11:30 a.m. 

This tutorial will equip institutional IT leaders and engineers with an understanding of strategies and solutions for supporting data-intensive research and post-secondary education. In contrast to enterprise solutions for administrative IT and generalizable “classroom” technologies, solutions for research — and for the teaching of research and data science methods — require solutions that vary by research project or course content, as well as uptime/redundancy requirements, access patterns, and the availability of federally-funded, off-campus resources.

Thus, the community-informed planning strategies and technology solutions covered in this tutorial will include overviews of available community-operated computing and data resources, the campus infrastructure needed to support access to those resources, and additional campus-level investments that bring advantageous capabilities while maximizing external funding opportunities to minimize costs. Along the way, we’ll address AI use cases, explore federal funding for campus cyberinfrastructure investments, engage participants in breakout discussions and hands-on strategic planning exercises, and point to later TechEx sessions discussing research-enabling IT infrastructure.

We recommend this tutorial for any TechEx attendee working at or with research and education organizations, including IT and research leaders, as well as those who operate and support information technology, from across research and education institutions, regional and national providers, and industry.

The Identity Problem You're Already Having: Practical Strategies for Improving Higher Ed Identity Proofing | 8–11:30 a.m.

Identity proofing failures surface as a compromised help desk account, a fraudulent financial aid applicant, a researcher who can't meet a federal compliance deadline. The expertise exists on campus. Understanding where it is and coordinating a campus-wide strategy is another matter.

This interactive workshop brings the problem of identity proofing into focus through a campus case study, peer table discussions, and hands-on activities. You’ll define the problem your institution is actually facing, identify the stakeholders who need to be involved, and begin a realistic action plan for moving forward. You’ll leave knowing what identity proofing means for your specific use case, whether that’s step-up authentication for research, fraud prevention in admissions and financial aid, or hardening help desk verification and what your next steps should be.

Barn Raising: Bring your own Challenge | 8 a.m.–4:30 p.m.

Bring your technical challenges to this collaborative, hands-on tutorial where you’ll tackle real-world problems with AI-powered coding tools and AWS expert guidance. Whether you’re managing technical debt, implementing coding standards, optimizing cloud workloads, automating infrastructure updates, or building enterprise system integrations, this tutorial provides the tools and support to make meaningful progress on your specific challenge.

Participants will work in their own AWS environments using agentic coding services—Kiro and AWS Transform—to address their unique problems. AWS Solutions Architects will provide foundational instruction on these AI-assisted development tools and offer individualized support throughout the day-long session. The format emphasizes peer learning: participants will share their challenges at the outset, collaborate and problem-solve throughout the day, and demonstrate their solutions or insights gained during structured report-outs.

This isn’t a traditional, follow-along tutorial. You’ll leave with working code, practical experience applying AI coding assistants to institutional challenges, and a network of peers who’ve tackled similar problems. Come prepared with a specific challenge, access to your AWS environment, and a willingness to learn from both experts and fellow participants in this barn-raising approach to cloud development.

Prerequisites: AWS account with appropriate permissions to deploy resources related to your specific challenge (e.g., EC2, Lambda, or relevant AWS services). Tutorial organizers will be in touch with participants in advance to provide more specific guidance.

By the end of this tutorial, participants will:
 

• Apply AI-powered coding tools to real-world challenges
• Implement automated code modernization strategies
• Leverage peer collaboration for problem-solving
• Develop strategies for streamlining workloads, enforcing code standards, and automating updates across their cloud estate

Beyond the Core: Deploying IPv6 to Users | 8 a.m.–4:30 p.m.

This full-day tutorial provides network engineers with both theoretical knowledge and hands-on experience in implementing IPv6 specifically for client networks. While traditional IPv6 training has focused primarily on core network deployment, this tutorial addresses the unique challenges and opportunities presented when implementing IPv6 in client networks.

The first half of the tutorial covers essential IPv6 concepts, including IPv6 basics, address planning, and a review of deployment options for end-user networks. We explore the spectrum from dual-stack implementations to IPv6-mostly and IPv6-only architectures, with special attention to transition technologies such as NAT64, DNS64, PREF64, and 464XLAT. We also cover technologies used to secure IPv6 Networks like RA Guard and DHCPv6 Snooping.

The second half of the tutorial shifts to virtual lab exercises where participants will configure NAT64, DNS64, PREF64, 464XLAT, DHCPv6-PD, as well as layer 2 protections like RA Guard and DHCPv6 Snooping. These practical tasks allow engineers to immediately apply theoretical concepts in a controlled environment.

Participants will leave with refreshed IPv6 knowledge specifically tailored to client-facing networks, practical configuration experience, and the confidence to implement secure and effective IPv6 solutions for end-user environments.

BGP and Route Policy for R&E Networks | 1–4:30 p.m.

Good BGP route policy is essential for routing security and to ensure your network traffic is making the best use of your R&E network connectivity and peering. This tutorial will cover creating a complete BGP routing policy for your network and best practices for configuring BGP. We will also look at useful tools for troubleshooting BGP routing issues.

eAC Cloud RADIUS Community Dialogue | 1–4:30 p.m.

Members of the eduroam Advisory Committee’s Cloud Infrastructure Working Group are hosting a half-day event to discuss the benefits and challenges of moving to a cloud-based solution. Community members and industry representatives are invited to engage on technical and business requirements.

GitOps with ArgoCD – Declarative Kubernetes Cluster Management | 1–4:30 p.m.

Modern Kubernetes deployments benefit significantly from a GitOps-based approach, where Git serves as the single source of truth for declarative infrastructure and applications. This half-day hands-on tutorial introduces participants to GitOps principles and practices using ArgoCD, a popular declarative continuous delivery tool for Kubernetes.

Designed for conference attendees with little to no Kubernetes experience, this tutorial will guide participants on adopting GitOps for cluster management. Attendees will start by understanding core GitOps concepts and how they solve common challenges in traditional deployment workflows. They will then install and configure ArgoCD in a pre-configured Kubernetes environment, learning how to connect Git repositories as the source of truth for their cluster state.

Learning from hands-on exercises, participants will deploy sample applications using ArgoCD, configure automated sync policies, and implement multi-environment deployments across development, staging, and production namespaces. Attendees will experience the advantages of declarative configuration management by making changes through Git commits and watching ArgoCD automatically reconcile the cluster state.

By the end of this tutorial, participants will be able to:
- Understand GitOps principles and benefits for Kubernetes cluster management
- Install and configure ArgoCD in a Kubernetes cluster
- Connect Git repositories to ArgoCD and manage applications declaratively
- Deploy and manage applications using ArgoCD
- Implement automated sync policies and manual approval workflows
- Configure multi-environment deployments with environment-specific configurations
- Perform application rollbacks and manage deployment history

Prerequisites
- Familiarity with YAML configuration files
- Basic Git knowledge (commits, branches, pull/push)
- Laptop with a web browser and terminal access
- Pre-configured Kubernetes environments will be provided

Audience
- DevOps engineers and platform teams looking to adopt GitOps practices
- System administrators managing Kubernetes clusters
- Developers interested in modern deployment workflows
- IT professionals transitioning to cloud-native infrastructure
- Anyone seeking to implement declarative, version-controlled cluster management

Grouper MCP – Yeah you know me (AI) | 1–4:30 p.m.

Grouper has been adding AI - MCP functionality in Grouper for 8 months.  Or at least it will be 8 months by the time TechEx rolls around.  Once institutions have upgraded to v7 and are cleared to use AI for their Grouper deployment, they will accelerate their access management practice like never before.  If you are not going to read this whole abstract, read this part: YOU NEED CLAUDE CODE PRO ($20) FOR THE HANDS ON PART OF THIS SESSION.

In TechEx 2025 we had a relatively successful half-day session on AI and Grouper.  This one will be better.  It will be less technical.  It will work better.  It will be glorious.

We will cover in a hands-on session (or just observe):
 

1. Configuring Grouper to use MCP
2. Connecting to the MCP
3. Access levels
4. Scopes to limit MCP ability
5. Recipes
6. Institutional documentation
7. Use cases

Throughout this session we will highlight security controls.

Hands-On AI-Powered Network Troubleshooting with LibreChat and MCP | 1–4:30 p.m.

Network operations are increasingly complex, and Generative AI offers practical new approaches to diagnosing and resolving issues faster. In this half-day tutorial, attendees get hands-on experience using an AI-assisted troubleshooting environment built on LibreChat with an integrated Looking Glass MCP (Model Context Protocol) server — no prior AI experience required, just a laptop with a browser.

We’ll open with a concise introduction to how large language models work, how tool calling enables an LLM to query live network state, and how these capabilities come together in a real troubleshooting workflow. Attendees will then connect to a hosted environment and work through guided scenarios, querying the network and interpreting AI-generated analysis alongside traditional looking glass output.

In the second half, we raise the stakes: we’ll introduce real faults into the network and give attendees time to diagnose them using the tools and techniques covered earlier. This is where the learning solidifies.

Throughout the session, we’ll share context on what Internet2 is building toward — an open AIOps framework designed to augment network operations teams with AI-powered tooling. Attendees will leave with a concrete understanding of how LLM tool calling works in a network context, what an AI-assisted NOC workflow looks like in practice, and how to begin exploring these capabilities within their own organizations.

Prerequisites: A laptop with a modern browser and SSH client. VS Code or preferred IDE with remote SSH optional.

Hands On: Getting to Know COmanage Registry v5.0 | 1–4:30 p.m.

New to or curious about COmanage Registry? Planning a migration from Registry v4 to v5? This is your opportunity for a hands on tutorial on using Registry v5, led by the project architect and developers of the COmanage Project. During this half-day session, you will learn about how to use COmanage Registry v5 for lifecycle management of your populations. Topics include how to model your organization or collaboration within Registry; setting up enrollment flows to automate data processing and link account data; and manage identifiers, groups, and roles to prepare information to be used in your other systems.

This session will highlight exciting enhancements to previous Registry versions, however, no prior experience is required.

COmanage Registry, created by the COmanage Open Source Project, is an identity registry with flexible enrollment and lifecycle management capabilities that helps you meet your identity management objectives using standardized tools and approaches. It can be used as a central person registry, a guest management system, or a collaboration hub for scholarly collaborations.
 

Immutable Linux Operating Systems: An Opinionated Survey and Hands-on Lab | 1–4:30 p.m.

A survey of the state of "immutable" Linux operating system ecosystems will be discussed. Particularly: the Red Hat affiliated ecosystem (ostree, rpm-ostree, bootc), kairos and elemental-toolkit ecosystems, and hypothetical future ecosystems such as those proposed by Lennart Poettering of systemd fame. A strong attempt will be made to showcase the critical differences amongst the approaches and how those affect the effort required downstream to employ the built solution in a scale out fashion. The main focus will be on provisioning bare metal computing hardware as this remains a big challenge in IT operations and can be reasons organizations elect to continue licensing costly solutions such as vmware. The hope is for an interactive dialog during this phase to find out what challenges members have when it comes to bare metal ops. Challenges should be identified candidly with discussions occurring on how the community can support each other, and perhaps where improved coordination with external open source projects might be beneficial.

The second 1/2 to 2/3 of the session should be hands on and will allow participants to get hands on time building a "custom" immutable Linux OS and simulate deploying it onto a virtual machine (maybe we can do real metal?). Participants will be provided with skeleton projects in flavors both for ostree-based and kairos-based options to work from. Participants will need to bring a laptop that can either run Linux virtual machines or is running Linux as the host/bare metal OS.

The Spy Who Automated Me: Deploying Quick Agents Across Campus | 1–4:30 p.m.

Your mission, should you choose to accept it: Deploy intelligent agents across your institution to automate workflows, generate insights, and execute mission-critical tasks—all without breaking cover.

In this hands-on tutorial, you'll explore the fundamentals of Amazon Quick and discover how custom AI agents can infiltrate every corner of your campus to streamline operations and enhance productivity. Like a well-trained intelligence network, Quick Agents work autonomously behind the scenes—monitoring data sources, executing complex tasks, and reporting back with actionable intelligence.

Tutorial participants will create their own Quick Agents tailored to real institutional challenges: automating student support workflows, generating operational insights from disparate data sources, coordinating cross-departmental tasks, and more. You'll learn agent design principles, deployment strategies, and governance frameworks to ensure your agents remain under control (unlike certain fictional spies).

Whether you're running a one-agent operation or building an entire intelligence network, you'll leave with working agents, practical implementation strategies, and the confidence to deploy AI automation across your institution. No tuxedo required—but bring your real-world challenges and a sense of adventure.

Who should attend:
 

• IT leaders exploring AI automation opportunities 
• Student services professionals seeking to streamline support workflows 
• Research computing staff supporting faculty and researchers 
• Operations leaders looking to enhance institutional efficiency 
• Anyone curious about deploying AI agents without extensive technical expertise 

Prerequisites:
 

• Laptop with web browser and AWS account access (instructions provided in advance)
• No coding experience required—this is a no-code/low-code tutorial
• Bring 1-2 real challenges from your institution that could benefit from automation

By the end of this tutorial, participants will be able to:
 

1. Design and deploy custom Quick Agents to automate specific workflows and tasks within their institutional context, using no-code/low-code approaches
2. Configure agent data sources, instructions, and integrations to connect with institutional systems and execute mission-critical operations autonomously
3. Implement governance frameworks and monitoring strategies to maintain control, ensure quality, and manage agent deployments at scale across campus
4. Evaluate use cases where AI agents can enhance productivity and streamline operations in student services, research support, and administrative functions
5. Understand an implementation roadmap for expanding from single-agent deployments to coordinated "intelligence networks" that work together across departments

Mobility Day at TechEX | Monday, Oct. 26 from 8:30–11:30 a.m.

Mobility Day is an annual workshop discussing topics including eduroam, mobile networks, 4G, 5G, WiFi 6(e), Internet of Things, eduroam for K12s, Low Power WAN, Passpoint/Hotspot 2.0 and other mobility-related topics.

Federation Day | Monday, Oct. 26 from 1:30–4:30 p.m.

This workshop shares current work that is ongoing within the InCommon Working Groups of CTAB and TAC, how it is progressing, and sharing (anticipated) substantial reports and proposals for standards for “Federation Expectations” aimed at use-cases like federated access for research and electronic resources, such as journals.

Intended audience includes Federation participants, IdP/SP operators, implementers, policy and trust framework stakeholders, and anyone involved in identity and access management in R&E.

We look forward to creating a highly interactive, participatory workshop that brings together many of the threads running through the identity infrastructure in our community today.

NetGurus | 8:30 a.m.–4:30 p.m.

NetGurus is a group of campus Network Engineers/Architects that meet to contribute and learn from each other for the betterment of the broader education and research community. Participants discuss networking topics in a round table format to encourage open discussion and knowledge sharing.

Typically, this group meets before or after a conference to discuss items of interest. Many times, topics are suggested ahead of time to encourage participation. To allow for orderly discussion and to maximize individual participation, meetings are limited to 30 attendees on a first come, first serve basis. Also, please limit participants to a max of two (2) per institution. NetGurus is a gathering of the senior network engineers and network architects for major universities. Attendance is restricted to those people.