Workshops

AI Workflows with ACCESS Pegasus

Karan Vahi and Mats Rynge

Workflows are a key technology for enabling complex scientific computations. They capture the interdependencies between processing steps in data analysis and simulation pipelines as well as the mechanisms to execute those steps reliably and efficiently. Workflows can capture complex processes, promote sharing and reuse, and also provide provenance information necessary for the verification of scientific results and scientific reproducibility.

Participants will learn to use Pegasus, a leading scientific workflow management system now integrated into the ACCESS Support offerings (https://support.access-ci.org/pegasus). ACCESS Pegasus provides a fully hosted environment built on Open OnDemand and Jupyter, enabling users to develop and run workflows directly from a web browser. Workflow execution is powered by HTCondor Annex, allowing jobs to run across multiple ACCESS resources, including PSC Bridges-2, SDSC Expanse, Purdue Anvil, NCSA Delta, and IU Jetstream2. Through hands-on exercises in a hosted Jupyter notebook, participants will work through an example LLM-RAG (Large Language Model - Retrieval Augmented Generation) workflow that leverages GPUs across ACCESS resources. Along the way, the tutorial will address key challenges and best practices across the entire workflow lifecycle.

Pegasus (https://pegasus.isi.edu) is used in a number of scientific domains doing production grade science. Pegasus is used in astronomy, gravitational wave science, bioinformatics, civil engineering, climate modeling, earthquake science, molecular dynamics and other complex analyses.

Agile Foundations for RSEs: Building an AI Assistant with Agile

Tisha Charles and David Luet

This 90-minute workshop introduces Agile methodology and its application to RSE projects. Participants will learn the core principles and values of Agile and explore common frameworks such as Scrum and Kanban. The workshop includes a hands-on exercise where participants collaboratively plan a simplified “Gemini Gem” feature, applying Agile techniques to user story creation, sprint planning, and backlog management. Google Gemini Gems are customizable versions of the Gemini AI chatbot, allowing users to create specialized AI assistants for specific tasks. This practical approach will equip RSEs with valuable skills to manage complex software projects, foster collaboration with researchers, and deliver incremental value in research. No prior experience in Agile or Gemini Gems development is required. Participants should bring a laptop, have a Google account that has access to Gemini, and, optionally, a GitHub account.

Learning to Talk with Your Users: User Experience Research for RSEs

Johanna Cohoon, Rajshree Deshmukh, and Mary Goldman

A few conversations with users can have a big impact. User research helps engineers uncover unmet needs, design more intuitive tools, reduce development churn by catching issues before they’re coded, and decrease support needs—ultimately saving time, effort, and cost across the project lifecycle. This workshop will provide practical instruction on a common technique for user research: semi-structured interviews. Semi-structured interviews are a flexible method for gathering rich information about experiences, perspectives, values, and challenges that can be used to improve research software products. Workshop participants will learn about recruitment, protocol development, and analysis and will leave the workshop with reusable resources for future use. Through this training, research software engineers will gain confidence in identifying user requirements and experience issues, gathering actionable feedback, and prioritizing future development directions.

An Experimentalist Approach to Software Testing

Jeff Soules and Brian Ward

Most builders of research software believe they should write tests for their tools, yet many struggle to do so efficiently and effectively. Formal strategies like Test-Driven Design may seem intimidating or inaccessible, while developers relying on an ad-hoc approach can suffer from “blank page syndrome” and struggle to provide efficient, complete coverage of essential functionality. Sometimes this leads to tests which actually spend more time testing mathematical properties or theories rather than the implementation, wasting both compute time and developer time; in other cases, developers fall back to manual tests, leading to developer anxiety and questionable software reliability.

In this workshop, we introduce a scientist-friendly perspective on software tests: tests are controlled experiments run on the implementation code. We will demonstrate a minimal setup for automated testing and the features to expect from a good testing system, and cover examples of common failures in test design. Participants will have the opportunity to apply this approach by revising practical examples—both of flawed tests, and of implementation code that creates friction when writing tests—before we reconvene as a full group to discuss solutions.

A computer will be needed for full participation, though attendees without one will still benefit from the discussion. Exercises will be provided in Python and C++.

Elianna DeSota and Jon Starr

U.S.-funded data repositories are increasingly unstable, losing funding, or shutting down entirely without closing processes, or a clear path towards maintaining this data. This rupture not only threatens long-term access to research outputs, but they highlight how fragile our current models for data stewardship are.

This workshop aims to unlock the latent knowledge in the RSE community by crowdsourcing pre-existing solutions for storing and sustaining large datasets, identifying the barriers to use, and agreeing on one step we can take together towards sustainable data infrastructure. The RSE community is the current stewards of our research infrastructure, and are both most at risk when data infrastructure is shut down or deprioritized. We’re also most well-positioned to define solutions that can support continued research.

Computers are not required, but recommended for digital note taking.

Introduction to Multi-Animal Pose Tracking with SLEAP

Elizabeth Berrigan, Andrew Park, and Talmo Pereira

This hands-on workshop introduces SLEAP (Social LEAP Estimates Animal Poses), an open-source deep learning framework for pose estimation and tracking—the process of identifying and following grouped landmarks, represented as nodes and edges, across video frames. Originally developed for animal behavior research, SLEAP is now a general-purpose toolkit used in diverse applications including plant phenotyping, disease progression monitoring, reinforcement learning, and behavioral analysis in space.

Participants will learn the full SLEAP workflow: defining skeletons or graph structures, labeling keypoints, training and configuring deep learning models, running inference, tracking across frames, proofreading predictions, and exporting results. The workshop highlights SLEAP’s intuitive human-in-the-loop interface, fast training and inference (with support for pretrained models), and flexible Python API for integration into research pipelines. To support hands-on participation, each attendee will access a cloud-based virtual machine with SLEAP pre-installed and GPU acceleration. No prior deep learning experience or software installation is required—just a laptop with Google Chrome, a stable internet connection, and a Gmail account.

By the end of the session, participants will be equipped to apply SLEAP to their own datasets for efficient, scalable, and reproducible pose estimation and tracking.

Hands-on scalable cloud automation and platform services for AI using CyVerse

Edwin Skidmore, Sarah Roberts, and Michelle Yung

Ready to empower your research project or community of users with the latest AI infrastructure? This hands-on workshop provides Research Software Engineers (RSEs) with the skills to build and support your own bespoke, secure AI/ML infrastructure using the powerful resources of CyVerse.

As RSEs, we are increasingly being called upon to provide custom Large Language Model (LLM) and machine learning (ML) integrations to address research problems. Identifying a platform that is both developer-friendly and user-friendly for deploying modern applications – such as those built with Gradio and Streamlit – along with an ever-growing ecosystem of MLOps tools, while providing an integrated institutional authentication, can be daunting. Discover how CyVerse can help you deliver these software and services faster while providing access to high-performance computing, massive data storage, and flexible cloud environments – including commercial clouds (AWS, Azure, etc) and cutting-edge GPUs available at no cost through NSF Jetstream2!

We’ll guide you through using CyVerse’s CACAO (Cloud Automation & Continuous Analysis Orchestration) for effortless deployment in the cloud, including your very own LLMs. You’ll also learn how to harness CyVerse APIs to seamlessly integrate its services into your custom applications and scientific workflows. Bring your laptop, sign up for a free CyVerse account, and get ready for practical exercises that will get you started in building tailored tools that can converse directly with your research data!