NSF Support Video Transcript

[Starting slide – NSF Support of Cyberinfrastructure Ecosystem and CI Professionals’ Development: Tom Gulbransen, Program Officer, TGulbran@NSF.gov, 703-292-4211]

Elizabeth: Nancy Glenn, our vice president of research, but before I do that, I wanted to please ask everyone to eat breakfast. Eat two or three Bagels, and please fill your pockets.

Also, Jenny Fother Gill wanted me to draw your attention to the ether pad and make sure that you visit the luminary to see the poster competition and research data as Art Exhibit and give us your thoughts because that’s what we’ll use to judge the entries. We will have fifty-dollar gift certificates from Stem Trek non-profit for the winners and Amazon…

We have a hackathon that’s ongoing co-located, we have students registered from Rwanda and Botswana, as well as the Boise State Siam Student Chapter. So they will be doing their thing in a separate room. They may be coming in and out; hopefully, they’ll eat donuts and bagels, but they will also have Amazon credits as prizes and a 50-dollar gift certificates for the winning team of four. So cheer them on, and some are participating from afar in Botswana; they’re actually having a pajama party in the lab because they don’t have internet at home. So we had permission from the chair to open the lab all night for them.

So anyway, thank you so much, and please welcome Nancy Glenn, our vice president for research.

[Turned camera to Nancy]

Nancy: Thank you. Well, it’s wonderful to be here. Thank you all for joining, and thank you to the Research Computing team and Elizabeth for organizing the events and for all of you that are participating in the workshops and the different breakout sessions. So thank you very much.

I’m Nancy Glenn, and I am pleased to be able to introduce Director Gulbransen, and when Elizabeth sent me his bio, and Director Robinson is on Zoom with a little bit of a travel snafu, so we’ll have to welcome him to Boise State another time, but when I read his bio I was so excited because I’ve used many of the resources which he has directed both exceed as well as I’ve participated heavily with Neon and Boulder so I’ll get started here with enthusiasm.

Tom Gulbransen is a program officer at the National Science Foundation’s Office of Advanced Cyber Infrastructure. He serves in that capacity as a rotator on temporary assignment from his career as a large program manager at the Battelle Memorial Institute, where he co-founded their environmental data sciences practice many years ago.

Tom helps with a variety of activities as NSF as a rotator. His primary focus directs the ACCESS program, which is Advanced Cyber Infrastructure Coordination of Ecosystem Services and Support. ACCESS expands upon the Exceed program, which served high-performance computing researchers for nearly a decade. Another part of the ecosystem Tom helps with is NSF’s learning and Workforce Investments. Especially those aimed at developing cyberinfrastructure professionals. Those programs together amount to approximately 80 million in awards.

Prior to his current duties, Tom helped complete the construction and operation of the Cyberinfrastructure for NSF’s National Ecological Observation Network NEON, leading Boulder, Colorado. He also co-led the environmental cyberinfrastructure for BP in their response to the Deepwater Horizon disaster.

I am so very much pleased to welcome Director Gulbransen over to you.

[Camera switched to Tom Gulbransen]

Gulbransen: Thank you, Nancy. Really glad to be here. I was able to listen in on some of the sessions this morning, but I wanted to physically be with you. I’m really sorry that we couldn’t get a connection… the flight from Salt Lake last night was just canceled, and everything was booked this morning, and Boise is popular, so I couldn’t sneak in. I have to acknowledge, though, that Elizabeth offered to drive all the way to Salt Lake to pick me up, and I didn’t take her up on that offer, but I really appreciate her trying to make it happen.

So let me spend what time we can. I’ll speak for maybe so 35 or 40 minutes, and we can have questions if you’d like; Jenny will have to catch them in the chat, though, because I won’t be able to watch for chat Q&A kind of things, but I would be willing to stop at any point and speak, and I really wanted to be able to spend time at lunch and today and tomorrow with you so please do write that phone number down or the email. I am very open to one-on-one conversations or if a group wants to get together and talk; my job at NSF is in official capacities; it’s to do all these grants and award dollars and be, but I would mostly like to just be able to listen to and learn what y’all need and want to do better because we can make some changes at NSF.

So today’s message to you is that I’m trying to listen and that our programs are willing to evolve as we can. So if I can make my slides move forward. Can you hear me, okay?

[Went to next slide – NSF Office of Advanced Cyberinfrastructure: Foster a cyberinfrastructure ecosystem to transform science and engineering research… through Research CI and CI research. List of workers: Manish Parashar(Office Director), Amy Walton(Deputy Office Director), Bob Chadduck, Ed Walker, Andrey Kanaev, Kevin Thompson, Bill Miller, Martin Halbery, Bogdan Mihaila, Robery Beverly, Tom Gulbransen, Alejandro Suarex, Seung-Jong(Jay) Park, Varun Chandola, Amy Apon, Ashok Srinivasan, Daniel Bullock, Juan (Jen) Li, Sharmistha Bagchi-Sen, Sharon Brode-Gave. “We’re hiring!” message below office directors and above the rest of the workers.]

I’ll just transition to the next slide. There you go. I’ll start at like the 36,000-foot level, even though I didn’t get a plane, and just start at the NSF. How is it structured, and then get into exactly what we can do today and going forward.

NSF, generally speaking, is leading progress in science across all states and territories plus International collaborations towards National Health, Prosperity welfare, and even National Defense. So how does that happen? The NSF leaders listen to the White House and the whole office of science technology policy in Congress and state representatives and University leadership consortia industry talks to us, and so we’re trying to listen to a lot of different signals, and that is helpful at the Grand largest scale but really, if it’s the eight directorates.

The director Im is dominant cyberinfrastructure information systems engineer. In the office that I’m in specifically, we have programs. These programs are like mid-scale research infrastructure or advanced computing networking and cyber infrastructure and campus cyberinfrastructure. I’m sure you all are aware of that. The ones I want to zoom in on today are about the organization of these programs in particularly how it relates to software and tools and organizations, and then the cyber construction professionals called Workforce, but the people is really what I’m into.

My personal experience has been in software development decision or support systems, and then I kind of helped teams to do it. So I don’t code as much anymore, but I did back some time ago, with some early stages of inference development and lots of database management and data sciences-based work. So I am very respectful of how much easier things have gotten in containerization and Globus movement of data across the globe.

[Went to next slide – OAC Mission: Foster a cyberinfrastructure ecosystem to transform science and engineering through Research CI and CI research, People, organizations, and communities, ACCESS(Coordination & User support), Gateways, Hubs, and Services, data infrastructure, software and workflow systems, CI-enabled Instrumentation, Pilots & testbeds, computing resources, R&E networks, security layers, and cloud resources & services.]

Let me show you again this broad-scale picture of what OAC US thinks about in a mental model and then how we’re tieing it together and implementing programs that you all contribute to or we would like to have you contribute more. We see things from a bare metal point of view, from a software development point of view. Now the cloud is offering interesting options, and then there are these intermediary roles that is vitally important; these gateways are based on certain domains or communities. How does that all happen?

[Clicked on slide show – diagram of observations and discoveries appeared: data infrastructure -> gateways, hubs, and services -> public access, open science -> pilots & testbeds -> software and workflow systems -> CI-enabled instrumentation -> R&E networks, security layers -> computing resources & services. The message below – “Rapid (disruptive) changes in S&E, users and CI landscapes -> Cyberinfrastrucutre ecosystem must evolve!”]

It’s the connectivity from raw observations through to Discovery. We’re trying to help science, and it’s really all about science. From two perspectives, one is we’re pushing forward to help research cyber structure and to help research the infrastructure itself, so we’re going to go in both directions.

[Slideshow transitioned with text above and diagram shrunk – “Foster a cyberinfrastructure ecosystem to transform science and engineering research… through Research CI and CI research”]

that turns into

[Went to next slide – OAC Vision: NSF’s vision for a National Cyberinfrastru future Ecosystem for Science and Engineering in the 21st Century | An equitable, agile. Integrated, robust trustworthy, and sustainable CI ecosystem that dries new thinking and transformative discoveries in all areas of S&E research and education | Overarching principles: View CI more holistically, Support translational research, Balance innovation with stability, Couple discovery and CI innovation cycle, Improve usability; ensure equitable access | http://go.usa.gov/xm8bU]

This one big, you know, Vision. A little vision, mission, and objectives kind of thing. Agile is a word that we’re used to mean it’s going to keep evolving and iterating and changing, and robust means that it’s got to be reliable, and trustworthy with regard to security, for sure. Sustainability means we’re not trying to fund everything forever; we’re trying mostly to fund the initiation and the building of communities that can then have momentum and continue through various kinds of partnerships that might not even involve NSF money forever but that the first word is the one that I need to drive.

Primarily it’s equitable; you’ll hear me say, democratization. This is one of the push points that we’re really trying to find new ways to do. We have seen exceed, for example, in Terror-Grid before that phenomenal accomplishment in science work together with the covid consortium in a week’s time where we had researchers around the globe moving data sharing models and jumping across systems that previously hadn’t connected together. That’s agility. That’s a great thing. It was done, and it has a Terror Grid and exceeds showed a lot of phenomenal success at the centers at the power centers and the r1 that are leading in some of these areas, and we think that’s wonderful.

Now we want to try to find a way to get even more involvement or further involvement; we want to have this opportunity to be in more places than just those main centers. So we want to balance the stability and continuity of services that we’ve seen and that have worked so well, along with new ways to do this, new workflows, new networking, and new data assets that are going to come from different places or from the edges. So we have to be stable but also open to novelty, and the tension there is healthy and interesting to try to balance.

So let me

[Went to next slide – Broad, FAIR, and equitable access to a … is essential to democratizing science: Democratizing Sciece through Cyber-infrastrutuce | Signification barriers – Knowledge: Awareness, discovery, expertise, support – Technical: Allocation, access, on-ramps – Social: Awareness of the importance of access to CI, rewards structures | Complex tradeoffs/optimizations. News article on “Democratizing Science through Advanced Cyberinfrastructure also included in the slideshow.]

highlight a couple more things about why we’re trying to do this and who we listen to. Our director Manish Parashar put out a piece really recently, which shows we’re recognizing some of the barriers that were as simple as how do I get a request in and then be allocated time or space in our systems. Once I have an allocation, how do I make sure it’s working well? Who can I get help from? That’s where we’re seeing a lot of change, and maybe you’ll hear about some community feedback that’s saying, hey, wait a minute, it changed. That’s not a good thing. Well we hope it is a good thing as we continue to improve it. So that’s what we’re trying to do. Is to balance these changes going forward.

[Went to next slide – Realizing an Advanced CI Ecosystem for ALL: – Integrated and user-friendly portals and gateways for discovering and accessing resources – Access to local CI resources as part of a shared fabric of national CI resources reachable through high-speed frictionless data networking – Diverse and flexible allocation and access modes that support a diversity of users and applications – Agil, easily accessible, and scalable networks of experts providing embedded expertise and support that is responsive to local needs – Broadly accessible training targeting the spectrum of CI users and skills | An article – “The Missing Millions”: Democratizing Computation and Data to Bridge Digital Divides and Increase Access to Science for Underrepresented Communities (A. Blatecky, EADER) | https://www.rti.org/publications/missing-millions/fulltext.pdf]

The Missing Millions Report that Dr. Pledecky and many other folks contributed to showing us that it’s not just about allegations. Although there are other barriers. Some of it is skill based, and some of it’s not really skill-based. Some of it’s connectivity. So the office is putting out a lot of investments in networking between groups, but we want to see more of the power that happens when you know Kyle Shannon, for example, was describing containerization, then he just had this calm, assuring way of saying this will work, this can work and his ability to deliver those opportunities to run code delivering that to an ecologist or a physicist who has a scientific question in mind but is not quite sure how to do it. That’s the connection that we want. Democratization means we want to let Kyle and other cyber professionals get to and join interests with scientists who might not be on the same campus. Might not be in the same state. That’s what we’re trying to do is to enable more of that to happen.

[Went to next slide – NSF-funced Advanced CI Ecosystem: Leadership-class(navy blue stars): Frontera & U of Texas, Austin | Capacity Systems(golden stars): Anvil, Purdue University, Bridges 2, Carnegie-Mellon University, Delta, UIUC, Expanse, U of California, San Diego, Jetstream 2, U Indiana + partners, Stampede 2, U of Texas, Austin | Innovation Prototypes/Testbeds(light green stars): Neocorte, Carnegie-Mellon University, Voyager, U of California, San Diego, Ookami, Stonybrook University, NRP, U of California, San Diego, ACES, Texas A&M University | Cloud resources(light blue stars): Cloudbank, U of California, San Diego, CloudLab, University of Utah, Chameleo, University of Chicago | Distributed Services(purple stars): PATH/OSG, U Wisconsin, Madison, ACCESS, and Fourteen Partners.]

So back to the systems themselves, just to make sure we have the context in mind. These stars are just some of the major centers where we have the capacity, and some of this, for example, that nearest Texas, Austin, is still a few years out, but it’s the leadership class that’s going to be the fastest, certainly the fastest, for AI applications or capacity systems are there are just six here listed, and then some of the prototypes that are coming through these are just a few of the many that are either in test bed, soon to be operational, or are some of them are even evolving forward. Cloud resources are coming to the front and quite available through industry collaborations or Industry Partnerships, let’s say. Then finally, the systems are made available through these Services, particularly the open science grid or open science pool path, if you will, and then the ACCESS system. These are two really large programs that are all about making connections between people and software, and that’s what I want to spend some of our time doing.

[Went to next slide – Involving National Research Ecosystem: Future advanced computing ecosystem(Integrating data, compute, software, and educational resources for science and engineering leading with NITRD the National Strategic Computing Reserve) | National AI research resource(Integrating data, compute, software, and educational resources for the AI community) | Subcommittee on open science(Establishing an open data commons) | NITRD SC and IWGs National Discovery Cloud(Advancing research on the technical foundations of a discovery cloud environment)]

That’s what’s happening so far. This is what’s coming over the horizon. I won’t spend any time here. Just the message here is that program officers like me are trying to get funding for the flow and get teams to be pass their reviews and keep going each year. At the same time, we’re also trying to figure out the five-year agenda for solicitations that are upcoming so it’s exciting to see some of the most recent work, the Knighter D work that’s gonna probably lay out a national discovery cloud, and certainly, the national AI research resource that just published its report last month, two months ago now. That Nair is a large driving force that’s going to be enabling us to make a case for investment in even more systems and teams of people to make use of them. So that’s the large scale.

[Went to next slide – Exemplar Acceleration of Scientific Discover Through AI and Computational Thinking: Arrow pointing(Inside arrow – Computational thinking already an integral part of every scientific discipline: Sophisticated and high precision techniques for modeling, simulation, testing analysis, reasoning and predication, reliable and reproducible results, enables knowledge sharing and interoperability, enables systems thinking) towards info bubble(Inside info-bubble – Opportunities for Expansion: Reasoning(Artificial Intelligence for pattern discovery and generating scientific hypothesis), Data and Modeling(Digital twins for real-time modeling and accurate forecasting, Sensing and Analytics(Smart technologies for real-time data collection and aggregation), Verifying(Formal methods and theorem provers for checking mathematical proofs, hypotheses)).]

Let me just do one moment on an example. It’s tough to read this, but I’ll try to get through it. This is the marriage that we wanna help happen more often. Computational thinking is already happening, and modeling is getting more complex. Data connectivity is driving new thinking. That turns into “What do we want next?” What could we do next? So sometimes it’s a push for an idea… I need this better, and other times it’s you know what I can do for you? That back-and-forth is what we want to see more of. So I’m going to grab this ecosystem concept and just do a riff on it for a couple of slides here. I want to make sure I stay roughly on time. The ecosystem concept

[Went to next slide – Ecosystem Goods & Services: Recruiting, Providing, Transforming, Exchanging, Assimilating | Segmented Cycle Diagram with bubbles including Microclimate, Airchemistry, Soil-Biogeochemistry, Watercycle, Ecosystem-Strucutre, and Physiology(Microclimat transitions to Airchemistry through radiation/temperature, Soil-Biogeo-chemistry through soil/temperature, Physiology through radiation/temperature, and Watercycle through radiation/temperature | Airchemistry transitions to Physiology through emission, respiration, and gas concentrations | Soil-Biogeo-chemistry transitions to Airchemistry through emissions/respiration, and Physiology through nutrient availability | Watercycle transitions to Soil-Biogeo-chemistry through water content, and Physiology through water availability | Ecosystem transitions to Watercycle through root distribution, Microclimate through shading, and Physiology through allometric relationships | Physiology transitions to Airchemistry through emission/respiration, Soil-Biogeo-chemistry through litterfall/nutrient uptake, Watercycle through water uptake, and Ecosystem-Structure through biomass growth).]

in nature is about goods and services. Whether it’s organisms consuming nutrients, water, sun temperature, or energy, right? This is the way these systems work, and they’re

[Went to next slide – Ecosystem Goods & Services: Includes the same diagram with keywords surrounding the image(Cyclical, Resilient,Evolving Places, Communities, Metabolism, Organisms, Microbes, Biological, Chemical, Geological, Physical, C:N:P, Compounds, Food, Biomass, E)]

from a physical point of view, cycling things from carbon, nitrogen phosphorus up to compounds to food to biomass, and who’s doing it? Microbes organism’s metabolism occurs, and then communities that are structured from the bottom of the physical raw materials up through into ecosystems, and as climate changes, they change too.

[Clicked on slide show adding new keywords next to the existing previous text(Resilient, Evolving, Organisations, Architectures, Compute, Algorithm, Code/Containers, Storage, Security, Networks, Software, Hardware, Data, Information, Models, Knowledge)]

That’s exactly the same concept as what we’re talking about in our cyberinfrastructure ecosystem. We have raw data observations; things get put together models, get populated or run and yield insights based on code that goes through and does computation and has architecturally nowadays architecture that has code built into it; in some cases. That’s, I think, a fair analogy to this notion of give and take and codependency between… so what colleges do or try to figure out what the limiting factor is? What’s causing this, too… what could make it better, or what could make it more resilient if something goes wrong or if something changes?

[Went to next slide – segmented cycle diagram changed: bubbles now include Institute facing, Software facing, Systems facing, Data facing, Researcher facing, Workforce facing, and ACCESS(All bubbles connect to ACCESS(other than Institute facing), and ACCESS connects to all bubbles).]

Well, that’s already being figured out by virtue of, you know, Carc and Consortium. Who is saying, “you have to be systems-facing if you want to work this out.” Other people say, “Yeah, sure,” but you have to be data-facing in order to really know which numbers to get. Well, you have to be a researcher facing to figure out those different facings are exactly why the ecosystem concept, some of us think, is a viable way to try to figure out what to do. I would add here the one on the top maybe Nancy is still here with us. Some of what we’re seeing is the need or an opportunity for academic institutions to restructure perhaps a little bit about how some of the really special talents the assistant professionals; some call them unicorns. Thrive on a project, and then when it’s over. You have to look around and figure out what’s my next project, or they jump to a chemistry project, and then they have to help somebody in astrophysics. Jumping around is really suitable for some people who love to jump around, but it’s not as good for others. So we think there’s a need to try to establish a little bit more clarity in the career pathways to help folks stick around with this phenomenally valuable service that they provide. So here’s the ACCESS

[Went to next slide – Advanced Cyberinfrastructure Coordination Ecosystem of Services & Support (ACCESS): flowchart circumferencing a cycle diagram – flowchart(Engage Users, Allocate, Coordinate & Communicate, Measure, and Operate): Engage Users connects with Operate Integrate on the right and transitions to Allocate on the left through people, organizations, & communities | Operate transitions to Measure through data & software services  | Measure transitions to Coordinate & Communicate through Networking & Cybersecurity Services | Coordinate & Communicate transitions to Allocate through Advanced Cyberinfrastructure Resources & Services | Allocate transitions to Engage Users through People, organizations & communities) and cycle diagram(Therorize -> Observe -> Hypothesize -> Experiment -> and Anzlye) | Text at bottom(“Cyberinfrastructure Ecosystem = compute+data+SW+networks+experts+Coorditiona = Allocate + Support + Integrate + Measure + Communicate” | Link at bottom(https://ACCESS-CI.org)]

and I didn’t build that acronym, but I think it works to some degree. What I want to do is to go through this ACCESS program a little bit in detail. This is the primary role that I have had at the foundation for a couple of years now. We have many teams doing many different things mostly; these purple centers are services that connect together in various ways to help this ecosystem function.

Starting at the left-hand side allocations, you have to be given access to these systems. How do you know which system to use? Well, if you had a tool that you could look at and say, “Who has code like mine, where has that code run, and how well does it run on different platforms” that would be helpful to some people who are trying to get allocation proposals together. It turns out that the measurement services are putting together a pilot right now that will be available maybe next year… within the year. It will help do that simulation, and then it’ll also do some performance optimization or give you data to do your tuning. So those are examples of some of the services that ACCESS provides when someone has a question or has a system performance potential issue, there’s a whole use of support infrastructure being put together, and we’ll go over some of the details of that behind the scenes.

How does it all connect together? How do we have the ability to give somebody allocations to 14 to 16 different systems? We have identity management and secure means of authentication that allow that to occur. It also is set up so that while more users can get in, we also want more data types and more networking opportunities to be part of this ecosystem as well, so the integration team has these road maps that they’ve put together. It’s a method for describing what your system does, what your on-prem stuff does, and how it could conceivably be added to the ecosystem for use a lot like the way the open science pool has shares resource usage across.

We also have some coordination and communications. That’s not just telling the ACCESS providers what they’re doing with each other ACCESS Communications, and Outreach folks want to be amplifiers; if your institution has a program or an online hackathon, for example, we want to be able to give you a voice and give you access to even broader reach than you might have had otherwise. So communication is happening from an awareness point of view that’s part of our mission.

So that’s ACCESS without naming any names; let me

[Went to next slide – ACCESS Program Objectives:

• Science-driven program balancing needs of researchers, RPs, developers
• Maintain continuity of production-quality services and evolve as needed
• Balance stability with innovative pilots
• Learn, listen, and leverage network of NSF programs
• People-facing support existing PIs & democratize for new
• Systems-facing flexible use and integration of new resources
• Workforce-facing amplification of value propositions to institutions
• Recognition of Cuberinfrastruce Professionals (CIP) careers: Georgia Tech, NCAR, Boulder, University of Buffalo, MOHPCC, The University of Chicago, PSC, Indiana University, Tufts University, University of Kentucky, Ohio Supercomputing Center, University of Illinois, USC University of Southern California, UC San Diego]

just do a couple of name-drops. Down the bottom are all the primary institutions that have responsibility for one part of the services or another. Some people have multiple roles, but let me also say that ACCESS is the successor, as noted in the Exceed program. Exceed was one award that went to an awardee who had many subs awardees, and so, in some ways, ACCESS is a carry forward because it has been Awarded, but there are five of them.

Well, that’s not even the end of it. The five awards are sort of the primary services and then there are other Awards and other programs that the office has been putting together that are augmenting the ACCESS services, and I want to make sure that I try to convey that.

So here’s what the core Access core program is doing. Again always science-driven. We’re not doing it for the fun of it or to get one system more used than another. If the engineers and scientists need more of something, we are going to be one of the ways to identify whether there is sufficient capacity or not. The systems facing flexible use of new resources, we’re seeing that with some of the test beds, for example, Okami here in Stony Brook. They have local users, but it’s a pretty demanding and a special environment and so ACCESS is going to help bring more people into that and offer some of those specialists from the Okami community to connect with new users.

So you may have heard of the extensible computational support services from the Exceed program PCSS; it was a very successful program that would do a fairly large commitment program half a year-long, maybe sometimes even almost a year-long, effort, where Exceed specialists would help scientists do a project or an investigation or fulfill an award. They were, you know, maybe a hundred of those each year, and they were successful. ACCESS wants to do even more than that, and the difference is that ACCESS is going to do it with even more people; a broader range of folks will be given the opportunity to contribute and listen to and respond to needs.

So we have a collaboration that’s happening to the support services, which are now larger in scale than the ECSS program was exceeding. That hasn’t fully turned into phone calls and Slack conversations yet, but we’re now into our 10th month of the first year of the ACCESS program, and for the previous ten months, I haven’t been able to fully quantify what programs are being awarded and being budgeted, but I can tell you as of now, as of March of 2023, there are more funds flowing towards people who are providing cyberinfrastructure professional support than ever before and the scope will continue to grow as we have more Awards. I’ll show some of those programs in a little while.

[Went to next slide – Elements of the CI Ecosystem: ACCESS Program(CI Professionals, C&E Research, HPC Systems)]

As I mentioned before, the drive for ACCESS is to start with research, giving people access to these systems, and to do that with connectivity to the professionals. So there’s kind of like different tiers, and this is for better for worse; that’s the way a lot of the money flows out of NSF. We put systems out there, testbeds or production Quality Systems.Research is oriented based on the disciplines, whether it’s the SBE or astrophysics, or transportation and then cyberinfrastructure professionals are funded through networks and coalitions and awards for training the ACT. I’m sorry I went the wrong way. The ACCESS program

[Clicked on slideshow revealing connections and keywords for CI professionals, S&E Research, and HPC System. CI Professionals(CyberTraining, RCN:CIP, CyberTeams, RCD-Nexus), S&E Research(Domain PIs, AI Institutes, ACCORD), and HPC Systems(Gateways, Cloudlab, CloudBank, Pegasus, and OnDemand)]

is really the connectivity between those.

[Clicked on slideshow – New text revealed(SCIPE(examples)]

That’s what we’re trying to do. We are attempting to ensure that we’re breaking through the silos of excellence before and making it more across the organizational and independent of who was awarded the money. We want to let people be available to each other in other programs, and that’s true for the software as well. Gateways are a great example of it doesn’t matter where you’re from; I want to get you through and into the system; Open OnDemand and Pegasus are agnostic to systems. We want more people to know about that, and I think it was, yeah, Benjamin, who was describing some of that earlier in these on-demand efforts where you don’t need to know where the code is and containerization of the entire research program through a workflow in Pegasus. That’s Nirvana for some people. For others, it’s okay. You can go straight to the systems yourselves. We’re not disrupting that, but we’re trying to bring more people in through these additional paths.

[Went to next slide – ACCESS Year1 Achievements:

• Four types of allocation opportunities now from 1 page abstract to full proposals
• Graduate students may serve as PI of an allocation request
• Allocations marketplace introduced for transactions and dynamic pricing
• Open OnDemand, Pegasus -> Ask.CI -> Match+ -> Match Premiere
• CCEP network launched
• Identity and Access Management secure & flexible
• Integration Roadmaps & Concierge Integration Experts
• XDMoD Data Analytics Framework
• XDMoD integration with Open OnDemand]

What have we gotten? What have we achieved? What have we gotten through in just this first year?

Allocations used to be a quarterly process of a 15-page proposal that went through a review process that is still the way it works. Every six months it will be a panel that sits with and examines the merits of and the defensibility of a request for systems allocations for the maximum level, but there are three other levels now. One is just discovered and explored and then grow. So there are three levels that are driven by an abstract or a one-page proposal, or a roughly five-page proposal, and those requests are going into the ACCESS allocation team continuously with a, I’m gonna hesitate because I can’t remember exactly that number, with the expectation that you’ll get feedback in a short time. In like days or weeks versus having to wait for a semi-annual cycle. So we’re opening the door more often, and it can open more quickly, and it can open at the graduate student level. That’s a breakthrough that we heard and were able to allow. There’s still a need for a professor to acknowledge or associate, but the grad student can lead it.

There is a marketplace that’s beginning to come to forth, and that is a process where you’ll get an allocation for credits, or you know, cycles; even if you don’t know which machine would be the best for you, that request will be listened to and the different systems that are applicable can make themselves available to you and if a system particularly has a lot of capacity, they can actually discount their conversion rate, so you’ll get more time for the credits that you were awarded or less if they’re at capacity. So there’s a now it’s a back and forth. That’s in response to what the resource providers told us about how they needed a little more flexibility to make the best use of their capacity.

The Open OnDemand and Pegasus platforms are the most self-service-oriented ways that we are hoping the research community uses these tools, and then we sort of have different levels of help. The next one up would be sort of a chat session which will grow based on our own dialogues to find use cases and to find answers based on what’s already been solved or served in existing engagements, and then the match is a true matchmaking process, you have a question we have experts with that kind of skill, and there’s a system to make those connections. The most… the deepest engagements could actually involve a year of work together. ACCESS doesn’t fund those people, although we do fund travel grants to make people available; we expect that proposals would find somebody and the proposal would be submitted with that person built-in. So it’s a sort of bring your own funding; we’ll bring the expertise and match it together, and that actually has launched. So there’s a granting program and this support network that the match program is underway with growing.

I mentioned the analytics framework and the numbers that are being put together with Open OnDemand,

[Went to next slide – ACCESS Opportunities to Participate:

Governance

• • Eternal Advisory Board
  • Researcher Advisory Committee
  • Resources Provider Forum
  • Always recruiting reviewers
  • Affinity Groups

Performance

• • CSSN Community Engagement Program (CCEP)
  • Pegasus, OnDemand, MATCH-Plus, and MATCH-Premier Pilots
  • MATCH-Plus/MATCH-Premier Steering Committees
  • Collaborate with the Metrics team: XDMoD developers, beta testers

Community Building & Engagement Invitations Welcomed

• • HBCU Conf on Climate Change Tapia Conf, PERAC22, SC22, HPC-in-the-City Hackathon…

Professional development opportunities in place

• • Student Training & Engagement Program (STEP)]

So it’s been a hell of a first year. Internally, but then I think also our external presence. So here’s what I wanted to make sure I get to with regard to ACCESS. This is the opening, this is the welcoming part that we’re trying to get the message out about. Just like the way the foundation responds to what Congress and OSTP, and our advisory councils tell us. ACCESS is the same way; ACCESS has an external Advisory Board of about 6 to 16 people, maybe it’s 12 plus a few other appointees, and we have an advisory committee of researchers, and we have a forum for all the resource providers who we’re connecting people with and the allocation process in the beginning that decides about allocations and how much should be allocated.

That is a process of volunteers who are… we’re always in need of more volunteers, and I think we’ll In the next year or so, we might actually evolve that process a little bit to have allocation reviewers dedicated to certain areas. For example, AI Institutes are now communities of researchers, in some cases, 100 researchers who have needs that we might see allocations dedicated specifically to them. So the resource… the reviewer process is one that you’re welcome to be involved with.

The support group, Dr. Knutson and many other folks have a notion that in order to listen and figure out what’s needed next and figure out who’s good at providing the solutions, they’ll be forming Affinity groups, and I think we’ll see whether the AI concept sort of generically becomes a group or tool usage within certain models might be an affinity group, and that’s going to be another place to learn together or respond together.

I mentioned the community engagement program is CCEP; that’s the notion of essentially building a Rolodex of “Hey, I’m really good at this. I love to do this stuff. Does anybody need this kind of help?” That database of talent will be available and searchable and discoverable by folks who are looking for help. We also have community building, and we would like to learn more about places where we can hear about the scientists and the engineer’s needs. So part of what I was trying to do and getting to be with you would be to keep hearing more about what you have and the way it needs, and then finally, there’s a student training and the engagement program starts with about 15 recruits for three months, and then it gets down to nine for half a year and then a couple, three or four, end up doing a long-term run. So that step program will cycle a few times, and you’re welcome to take a look at that.

[Went to next slide – Recognizing People in the Ecosystem: CI Contributors(Cyber Scientists to develop new capabilities), CI Professionals(Professional Staff to deploy & support new capabilities), CI Users(Area Scientists to exploit new capabilities) all point towards OAC.]

Do I have enough? Yeah, let me just do a couple of minutes. I want to buzz through a couple of programs. This is the part about… how the Office of Advanced Cyberinfrastructure think about cyberinfrastructure professionals. We see professionals as different from people who are using the tools, and they’re different from the actual scientists

[Went to next slide – An advanced CI Ecosystem for All: User Support; Learning & Workforce Development | CI Professionals essential to democratizing CI…

• Scalable networks of experts providing embedded expertise and support that is responsive to local needs;
• Broadly accessible training and mentorships targeting the spectrum of CI users and skills.]

who are doing research. This is the professional deployed; they’re the folks that are going to actually help make a researcher get the job done,

[Clicked on slide – Revealed United States map with a color scale between 1 and 24 for white to blue]

and we have a few programs that have been put together and are funded and are going to be funded more to provide this the landscape of many networks and many groups of people that can be available, for example, the Cyber training program. There’s a research coordination network. The campus cyberinfrastructure program has that. There’s a minority serving “Inaudible.” that has been awarded. So we’re putting money out in a lot of different ways in order to provide many more opportunities for people to be ready and then for us to connect together.

[Went to next slide – Relevant CI-Related Programs: What will your project focus on? (Deleveop software or data repository, perform research that will enable future CI, support research that uses CI, foster CIP careers, provide training) | What gap will you fill? (Increase community CI for research or education, increase knowledge needed for CI, increases support for CIP career & research by underserved groups, increase research workforce to leverage CI) | What will your project deliver? (Community-sustained CI, techniques that will enable CI and a CI prototype, research support in CI, CIP career paths, scalable and sustainable training program) | First column is referred to as CSSI, the second column – is OAC CORE, the third column – is SCIPE, and the fourth column – is CyberTraining |  Text at bottom – Programs have specific purposes; however, are not necessarily mutually exclusive.]

It gets a little bit… I understand it’s hard to figure out all these different programs. I also struggle with some of the programs and how they’re different, but call a program officer, talk to somebody on NFS, and say, help me figure this out. That’s what our job is… to help you do that. So we have some of these little sort of cheat sheets that will help you to navigate “which program is the best place to try to seek funding for my idea.” If it’s cyber training, I have a role there to help figure out whether you’re material that you’re developing or the teaching that you’re doing, or the opportunities for learning that you’re doing are cyber training, or might it be that the SCIPE program is for strengthening the CI professional ecosystem? That’s one of the programs I’m doing, and here, it’s not so much about building training material; it’s more about readying the folks to go and be listening to and deployed to ACCESS and to other researchers on your campuses. So there are a lot of these programs that program officers

[Went to next slide – CyberTraining: Motivation:

• Advanced CI has a transformative impact o na charity of scientific research domains
• The research workforce will benefit from innovative disciple-appropriate training and curriculum materials
• There is a need to foster broad adoption of CI resources, tools, and methods by diverse research communities]

like me, who really wants to make it available to you. I’ll step really quickly through cyber training.

[Went to next slide – CyberTraining Solicitation Goals:

Long-term vision: Computational and Data-driven Science for All scientists and engineers

• • Prepare, nurture, and grow the scientific research workforce, including students, instructors, and research CI professionals

Ensure broad adoption of CI tools, methods, and resources

Integrate CI and CDS&E skills into undergraduate and graduate curricula

• • Address emerging needs and unaddressed bottlenecks through innovative and scalable training
  • Catalyze research with training and educational activities

Broaden CI access and adoption by varied institutions, scientific communities, and underrepresented groups.]

There’s the data-driven and science-driven again; this is for the adoption of tools and methods, so now we’re sort of proselytizing the assets and the skills. It can be on campus it

[Went to next slide – NSF-Wide Participation: Directorate(CISE(Divisions(OAC(lead), CCF, CNS, IIS), ENG(Divisions(CMMI and CBET)), GEO(Divisions(AII)), EDU(Divisions(DGE)), MPS(Dvisions(AST, CHE, DMR, and PHY), SBE(Divisions(SES)) | Text at bottom – OAC contacts: Ashok Srinivasan and Juan (Jenny) Li)]

can be multiple campuses, but here’s the way that it’s done really well. When you connect with somebody from Geo or from engineering or from education, and you have constituents that are saying I wish I knew how to do, and then you have people who are ready to solve that or to develop skills to solve that. That’s the best kind of proposition, and that’s where cyber training is that’s where the Merit review process, you know, sort of scores the most points.

[Went to next slide – Research and Education-related projects in the science/engineering domain:

• Pilot: Exploratory projects, $300k over two years
• Small implementation: $500K over four years
• Medium implementation: $1M over four years

1. Identify challenges in research workforce development
2. (a) Broaden use of CI resources (b) CI skills training – expected to coordinate with ACCESS (access-ci.org_
3. Scalability and sustainability of the training program
4. Recruitment and evaluation plans
5. Collective impact strategy
6. Fostering a suitable community

One through two are Pilot, one through five are Small, and one through six are Medium.]

At different scales, they can have really short exploratory ones, or it can go for longer time periods.

[Went to next slide – SCIPE: Motivations

• Strengthening ≡ , democratizing, connecting, recognizing
• CI ≡ Systems Software, Data, Services, Networking
• Developers, Users, Admins, Mentors, Computational Scientists
• Ecosystem of interdependencies & interfaces
• Leveraging existing program investments
• Expanding institutional recognition of careers]

The same is true for SCIPE. If we’re going to do democratizing, show what contributors and what people would like to do systems development and are ready but didn’t have access to the systems or to the researchers, show how you’re engaging with more people like that bringing them together and then deploying out to solve a particular kind of problem. That’s the same

[Went to next slide – NSF-Wide Participation: Directorate(Divisions(CISE(OAC (lead), CCF, CNS, and IIS)), ENG(Divisions(CMMI and CBET)), GEO(Divisoins(AGS, EAR, OCE, and OPP)), EDU(Divisions(DGE)), MPS(Divisions(AST, CHE, DMR, PHY)), SBE(Divisions(SES)), TIP(Divisions(RIE,PFI))) | Text at bottom – OAC contacts: Tom Gulbransen]

connectivity to other directorates; we like to see an acknowledgment from a program officer in a directorate as well as from our office, and if you don’t know who to call, call us, then we’ll help to make those connections to others.

[Went to next slide – SCIPE Program Context:

Funding:

• • Approximately $15M for up to 4 awards
  • Support for research CI professionals’ services
  • Up to 4 FTEs per year for up to 5 years
  • NSF directorates co-funding based on relevance: CISE, ENG, GEO, MPS, EHR, SBE, TIP
  • Longer-term plans requested for sustaining CI professionals
• Who are CIP:
  • Develop, manage, and support the effective use of research CI
  • Includes scientists, IT professionals, and engineers who research and develop new CI capabilities, approaches, and methods
  • Various facings: CI system administrators, CI research staff, research software engineers, data curators, CI facilitators, Computational scientists

Ideal for PIs who:

• • Want to accelerate the adoption of research CI
  • Are ready to connect & coordinate with S&E research communities
  • Seek to strengthen & broaden the diversity of the CIP workforce
  • Whose institutions will develop sustainable long-term career paths for CIPs]

So I’ll provide these slides so that you can have more of these details, but for example, in SCIPE, we will probably have… guess I can’t say how many awards, but multiple awards to teams of people who will have three or four full-time equivalent people available to serve science and engineering. At times a number of different awards don’t have to be four people. It can be fractions of people could be a dozen people that hear sharing on a part-time basis. That’s the kind of connectivity that we like to see done.

[Went to next slide – CyberTraining: Pilot: An Artificial Intelligence Bootcamp for PI: K. Tomko, CO-PIs: K. Cahill, E.fosler-Lussier, R.Machiraju, D. Panco | Training to enable connections by each facing of CIP roles |

Researcher Facing Role, Software / Data Facing Role, System Facing Role, Sponsor / Stakeholder Facing Role

Common foundation

• • • Research Use Cases
    • Introductory AI methods
    • Ethics for AI
    • Introductory AI frameworks & software environments
    • Data analysis workflows
    • Data management and privacy

Workflows & Tools

Business Models for AI

Researcher Facing Role

• • Workflows & Tools

Software / Data Facing Role and Systems Facing Role

• • HPC-Enabled AI
  • Performance & Scalability
  • Frameworks & Libraries
  • Configuration for AI

Sponsor / Stakeholder Facing Role

• • Business Models for AI

Additional CyberTraining: CIP awards in 2022…

1. Training & Developing a Research Computing & DataCIP Community PI: Thomas, CoPIs: Goetz, Sinkovitz, Rodriguez, Wagner (2230127)
2. Cross-InstitutionalResearch Engagement Network for CI Facilitators PI:Crosby (2230106)]

Examples of winners from last year… Karen Tomko put together a training linked specifically to the artificial intelligence tool sets and a boot camping process and has had a lot of success Nationwide, and enrollees from the across the nation going to their platform to learn and figure out what possibilities are.

[Went to next slide – CyberTraining: Pilot: A professional Development and Certificat Program for Cyberinfrastructure Facilitators (2118193) PI: Neeman, Co-PIs: Brunson, D. Colbry |

Why:

• • CI Facilitators training in mission-critical skills & proof of mastery

Who:

• • CI Facilitators across ecosystem at R1s, R2s, non-PhD-granting, MSI (estimated ~2000)

Solution:

• • Certified Cyberinfrastrcuture Facilitator Training & Development (CCIETD) Professional development certification Information Education (non-matriculated)

A badge for each skill

• • Professional/Interpersonal Skills
  • Understanding Researchers’ Circumstances
  • Technical Content
  • Research Data Lifecycle Management
  • CI Landscape]

Henry Neiman and others have been focusing on Mastery and the facilitator role, the ones that try to do that brokering of what you need. I think I know somebody who can help you with that from a leadership perspective, and then there are two others to note

[Went to next slide – Research Coordination Networks: CIP |

RCN:CIP: Connect.CI-based Community-wide Mentorship[ NEtwork (CCMNet)

• • PI: Brandt, CoPI: Christopherson, Brazil, Battelle, Gazula
  • Challenge: Keep abreast of the extensive array of tools, techniques, and specialized approaches needed to pursue today’s complex research questions.
  • Solution: Connect. CI-baed Community-wide Mentorship Network is developing a network for peer-to-peer knowledge sharing among subject matter experts, as well as mentor-centric relationships

RCN:CIP: Midwest Research Computing and Data Consortium (MWRCD)

• • • PI: Snapp-Childs, CoPI: Plale
    • Challenge: CIP responsiveness and innovation regardless of group size in highly dynamic environments of computationally-based research that encompasses new needs almost daily.
    • Solution: Follow a Community Participation Model to establish a regional network serving smaller, less well-resourced institutions to exchange insights, solve problems, and advocate.]

these research coordination networks that Kevin Brandt did in South Dakota, where he’s focused on community-based mentorships reaching folks that didn’t previously have availability, and then Winona Snap Childs has a Midwest geographically centered notion that within a day’s drive of each other, there should be some Community alliances and allegiances that would allow everybody to hop up above the campus level and trade-in between. This is exactly what we’re thinking is a way to break through and allow more of that flux, as I mentioned before, the exchange of ideas and the exchange of needs across the community.

[Went to next slide – Questions: Thank you for the opportunity to speak with you | TGulbran@NSF.gov | 703-292-4211]

46 minutes. Okay, can I try to take some questions, or what would be the most appropriate way to allow for some follow-through, or maybe you’re heading for lunch, I’d heard you mentioned bagels before and I’m a New Yorker, so I’m missing your bagels.

Elizabeth: Well, I can try to field questions. I’m getting a lot of echoing.

Gulbransen: Yeah, I can hear that.

[Switched camera to Elizabeth]

Elizabeth: You have not mentioned the campus Champion program.

[Switched camera to Tom]

Gulbransen: The campus Champions are best attached to the ACCESS program through that c step. The process of providing grants to people who want to try to compare notes like we’re going to have a branding program to get to perk this year, and so it’s through the use of support, and it’s foreign to the same kind of connectivity, but it’s not open-ended funding. It’s funding based on what the Campus-Championing is adding to the support team. If they have materials or if they are going to produce sessions they would get a grant of a certain amount of dollars, and Dr. Knuth’s team has… I can’t remember how many they put out, but they have put out a number of grants already, and they have an open process for vetting that. So yeah, Campus Champions are loved and admired, and I’ll be the first to admit they there isn’t enough written to re-establish how they can best work, so Dr Knuth and others are resetting those pathways. She was in the Campus Champions, Julie Ma has been a campus Champion, so they are both practitioners and leaders of the path; and I have a new system to get involved, and I know of other

[Switched camera to Elizabeth and back to Tom]

ideas that are being considered to kind of keep the campus Champions recognized and supported.

[Switched camera to Elizabeth]

Elizabeth: I think there are six campus Champions… raise your hand if you’re Campus Champion. Yeah, we’ve got five in the room today. One of the names on the paper you shared was Gazula Vikram Gazula was the first Campus Champion in 2008 from the University of Kentucky, and it grew to more than 300… so.

[Switched camera to Tom]

Gulbransen: Yeah, he’s pretty special.

[Switched camera to Elizabeth]

Elizabeth: He is, as our Henry and Dana

[Switched camera to Tom]

Gulbransen: Well, I think the special part of their personalities can’t be underestimated. That is part of it.; there’s a technical part, and then there’s a social part in it, and it’s wonderful to see folks like that prosper.

[Switched camera to audience member]

Elizabeth: It’s on, but it’s only working

Speaker (1): So just, I was curious about the status of the future of the EPSCoR program.

[Switched camera to Tom]

Gulbransen: I’m sorry that was a question about the EPSCoR program?

[Switched camera to audience member]

Speaker(1): Yes.

[Switched camera to Tom]

Gulbransen: I can’t give you very much detail about it other than to, you know, commit to following up. So if you could put your name in the chat, I’ll connect you with the right app score, folks. the way you pose the question, it sounded like you were concerned that it didn’t have a future, and I’m not aware of anything that would indicate that, so let me get some more details for you.

[Switched camera to Elizabeth]

Elizabeth: I noticed on the map of the United States there are no stars as far as capacity systems in the Pacific Northwest; are there plans to introduce more systems in the Pacific Northwest?

[Switched camera to Tom]

Gulbransen: There’s, I can only say that there’s a sensitivity to regional and geographic coverage, that’s for sure, and that’s from consideration that is promoted in the review process, but we don’t get to say, “Hey Northwest, we’re going to have a Northwest”… we can’t be that directive, but if a program were to be proposed, I think the geographic coverage would be favorable in that regard and I saw the same sort of lack of stars and dots in some of the other maps in the Idaho region, which is part of why I wanted to be there to be with you.

[Switched camera to Elizabeth]

Elizabeth: So we will think of another reason for you to come back and visit.

[Switched camera to Tom]

Gulbransen: Thank you.

[Switched camera to Elizabeth]

Elizabeth: That would be wonderful. Are there other questions? No? I have one more. As for the allocations with Terra-Grid, it was a service unit. What are the allocation increments under ACESS?

[Switched camera to Tom]

Gulbransen: The increments awarded carry forward from Exceed the difference now is the conversion rate at which they would be they would turn into cycles on a particular machine because that’s where the dynamic pricing could allow for more or less depending on capacity, and then there are other units or credits which are still to be figured out when it comes to space or data resources and trying to get at… it has to be scaled to the capacity you know what is the limiting capacity, and then what’s a fraction thereof that they can roll out? So some of the units are still being evaluated in the pilot program, but to get back to you to your earliest question, the SUs that carries forward, that’s still consistent.

[Switched camera to Elizabeth]

Elizabeth: Very good, thank you. Well, thank you for joining us today, Tom; we appreciate it, and we’ll bring you back again in person, so thank you.