High-performance computing (HPC) has become essential to modern science, but many researchers still face major barriers when accessing compute resources, collaborating across institutions, and managing scientific data workflows. Increasingly, research organizations are rethinking computing infrastructure not just as backend technology, but as a scientific instrument designed to accelerate discovery, reduce friction, and make advanced computing more accessible to researchers.
In this episode of Trends from the Trenches, BioTeam’s Jessica StLouis speaks with Moffitt Cancer Center’s Jarett DeAngelis and Shane Corder about how their team is redesigning research computing environments around accessibility, collaboration, usability, and scientific impact.
Key topics discussed
- Research computing and HPC accessibility
- Computing as a scientific instrument
- Reducing friction in scientific workflows
- Open OnDemand and self-service HPC
- Collaborative research infrastructure
- Scientific data sharing and authentication
- AI-ready research computing environments
- Researcher experience and workflow usability
Why Researchers Still Struggle With HPC
For many researchers, accessing high-performance computing environments still requires navigating command-line interfaces, job schedulers, authentication hurdles, data transfer workflows, and institution-specific onboarding processes.
As Jarett DeAngelis explains, collaboration across institutions often introduces additional operational complexity around permissions, data synchronization, identity management, and PHI controls.
“All of that kind of comes together to really create some massive amounts of friction when it comes to trying to collaborate with other people outside the organization.” — Jarett DeAngelis
At many institutions, researchers with strong scientific ideas may still struggle to effectively access the computing resources needed to execute them.
Computing as a Scientific Instrument
Moffitt Cancer Center recently received a $2 million NIH S10 grant to build what they call the Collaborative Computing Center, a dedicated computing environment designed to support collaborative scientific computing with outside organizations.
The project reflects a broader industry shift in how research organizations think about computing infrastructure.
“Computing resources very much are a scientific instrument.” — Jarett DeAngelis
Historically, grants of this type were often associated with purchasing physical scientific instruments like microscopes or imaging systems. Increasingly, however, advanced computing infrastructure is being recognized as equally essential to modern scientific discovery.
Applications, including bioinformatics, digital pathology, AI, machine learning, and large-scale imaging analysis, all increasingly depend on scalable computing environments.
Reducing Friction in Scientific Computing
A major theme throughout the discussion was reducing operational friction for researchers.
Rather than designing systems primarily around infrastructure administration, Moffitt’s Scientific Computing team is focused on improving usability, accessibility, and researcher experience.
“The idea here is to be as low friction, especially with respect to process, as possible.” — Jarett DeAngelis
The Collaborative Computing Center was intentionally architected to simplify onboarding and collaboration with outside organizations while maintaining appropriate security controls and data management capabilities.
The system includes:
- Dedicated compute and storage infrastructure
- Independent internet connectivity
- Flexible authentication models
- High-speed Hammerspace storage
- Integrated data lifecycle management
- Globus-based authentication and data sharing workflows
This architecture allows collaborators to securely access compute and data resources without requiring traditional institutional account provisioning workflows.
Making HPC Accessible Through the Browser
Another major focus of the discussion was Open OnDemand, a web-based HPC access platform designed to simplify how researchers interact with scientific computing systems.
Traditionally, researchers often needed to:
- Build Slurm submission scripts
- Navigate Linux command-line environments
- Configure SSH tunnels
- Manage software dependencies manually
- Learn institution-specific workflows
Open OnDemand abstracts much of that complexity away.
“Open OnDemand really breaks down the barriers.” — Shane Corder
Researchers can instead launch applications such as RStudio, MATLAB, VS Code, Galaxy, and terminal environments directly from a web browser via simplified interfaces.
“It puts us in a position to offer the power of HPC to pretty much anybody who knows how to use a web browser.” — Jarett DeAngelis
The platform also enables containerized application delivery using Apptainer, helping simplify software management and reproducibility across research environments.
Democratizing Access to Scientific Computing
One of the most compelling themes from the conversation was how these technologies are expanding access to advanced computing beyond traditional HPC power users.
Moffitt researchers are now building custom applications on top of Open OnDemand workflows, including tools that enable physician-scientists with little or no HPC experience to interact directly with advanced computational resources.
“It’s about helping researchers get their work done and do the science that they’re here for.” — Shane Corder
This shift represents an important evolution in research computing.
Rather than expecting researchers to adapt themselves to infrastructure, infrastructure is increasingly being adapted around scientific workflows and researcher usability.
The Future of AI-Ready Research Infrastructure
The conversation also explored how AI and machine learning are rapidly increasing demand for scalable, accessible computing infrastructure across cancer research and life sciences.
Moffitt’s Scientific Computing team is actively expanding local AI and ML inference capabilities to support:
- Bioinformatics workflows
- Digital pathology initiatives
- AI-assisted analysis
- Secure internal LLM workflows
- Large-scale scientific data processing
As AI workloads continue to expand, institutions increasingly need infrastructure that balances:
- performance
- usability
- scalability
- collaboration
- security
- reproducibility
The future of scientific computing may ultimately depend not only on raw compute power, but on how effectively institutions reduce friction between researchers and the infrastructure supporting discovery.
Questions explored in this episode
- Why is HPC still difficult for many researchers to access?
- How are research institutions simplifying scientific computing?
- What role does Open OnDemand play in reducing HPC barriers?
- Why is computing increasingly viewed as a scientific instrument?
- How can collaborative HPC environments accelerate cancer research?
- What does the future of AI-ready research infrastructure look like?
The full Trends from the Trenches podcast featuring Moffitt Cancer Center’s Scientific Computing team can be found here:
https://bioteam.net/podcasts/podcast-simplifying-hpc-resources/
