Space and Defense Innovation Launchpad Podcast - Episode 7

Brandon Frisco: Foreign.

Brian Fox: Welcome to the Space and Defense Innovation Launchpad podcast. I'm Brian Fox, and in this series you'll be hearing from different technologists and technology leaders in government, industry and academia about their efforts to implement innovative technologies to impact civilian and defense missions. Together, we will explore multiple aspects of this technical realm, including vehicle propulsion, material science, digital geospatial and sensor technologies. Hello and uh, thank you all for joining us in today's Space and Defense Innovation Launchpad. I'm, um, Brian Fox with the University of Tennessee Space Institute. And many thanks to ATARC for hosting today's podcast. This has been a lot of fun, um, resetting from digital modernization into the, the realm of space and defense. So many thanks as always, and again to ATARC for doing this. And I am so glad to have Gordon and Brandon joining us today. They're coming in from MTSI and we're going to be talking about data testing and evaluation. Um, and so it's really exciting. So, Gordon and Brandon, thank you so much for joining today. Um, I'll start with Gordon and then I'm going to hand it off to Brandon. Gordon, can you tell us a little bit about yourself, your organization and your role there? Sure.

Gordon Fogg: Um, my name is Gordon Fogg, and first of all, Brian, thank you very much for uh, for having us on here. Um, it's been enjoying, uh, enjoyment getting to know you and getting to know UTSI in terms of, uh, some of our common interest on the, uh, on the business side. Uh, um, you know, my background, I spent 22 years in the Navy. I was a tactical, uh, fighter pilot. Started out in F14s, ended up in, uh, F18s. Spent most of my time operational down, you know, downrange. And um, left the Navy in 2008 and came to Huntsville with MTSI. At that point was, um, supporting one of the organizations, uh, uh, on Redstone Arsenal. I'm still with that organization now, um, doing a variety of different things, mostly strategic test planning, test implementation in the, uh, at the project level and uh, within my company roles. I'm the business unit lead for the Missile Defense Services business unit. And uh, my primary areas is providing CETA support, uh, across the army and missile defense portfolio. And uh, also we have um, some work in our hypersonic test area as well, which, you know, I think a lot of the discussions we're going to have today will kind of, uh, pivot on, you know, scaling and testing and uh, how we get the most out of that across multiple stakeholders.

Brian Fox: Amen to that, Gordon. Brandon, over to you. Brief introduction Please, a little bit about yourself and your uh, the role at mtsi.

Brandon Frisco: Yeah, first of all thanks for having me and it's a pleasure to be here. Uh, so my name is Brandon Frisco. Uh, I've been in automation and data sciences for about 20 years now. Uh, my current role at the MTSI is the data as a data architect and uh, I am currently in my uh, doctoral dissertation research for AI machine learning. So I'm helping the uh, company uh forward those efforts. And uh, my background has been a lot in automation and computer, uh large scale enterprise computer systems. Uh, the goal is to bring kind of the contemporary paradigm to uh, industry from industry into the government sector and kind of unify those two things where we can have the meld of uh, the government space and the uh, private sector space. So a lot of my background has been in trying to integrate and implement a really hybrid approach to these things with many efforts. I've worked uh, uh, in all four areas of test training and tactics. So I've kind of seen a giant landscape of everything that's happened and uh, there is definitely avenues to do this and uh, many situations. But my focus is really trying to help that effort push it forward.

Brian Fox: Excellent. Yeah, thank you Brandon. Wonderful uh, to have you both on. Um, we are going to be getting into data, data interoperability all for really in the space of uh, testing and evaluation and what, what I've picked up since my time at UTSI starting last summer is whether it's folks in government, folks in industry or academia, there's a real interest to. There's been a pivot with the new administration deliver faster. Let's get these capabilities out there. But everyone is talking about the challenge of doing that successfully around testing that. Yep. We can't field until it's tested so we're going to be leaning in on that. Um, so to that, that like strategic national level interest to speed up testing. How is data sharing important to speeding up the development and testing of innovative flight vehicles? This could be missile defense missiles.

Brandon Frisco: Right.

Brian Fox: It could be sensors that are ground based or space based. Um, what's the role in data sharing for spill?

Gordon Fogg: I'll start out on that and then I'll let Brandon talk a little bit more about the details of how this um, could potentially work. But when you look at the reconciliation bill and the big beautiful bill that was passed and we'll kind of right now look at hypersonic testing because that was an area that was uh, called out. Right. We look at our peer competitors and the threats, um, and we see that you know, there needed to be a larger emphasis on the delivery of both uh, offensive hypersonic weapons as well as our ability to you know, counter what the threat is doing in that ah, particular regime. And you know, so like the government, when we see something that needs more emphasis, we put more money out there against it. Right. Um, and so you know, testing is a key aspect of that. And what ends up happening is you accelerate testing, you buy more test vehicles, you expand your range capabilities and what you end up doing is you generate a ton more test data. Um, that test data may be relevant to filling in knowledge gaps that we're trying to get in order to uh, build, develop and deliver capabilities to the warfighter or they may not be relevant. And uh, in addition to that, what you have across the hypersonic, uh, basically community is you have multiple different stakeholders. Um, and this really is something that I think creates a really big challenge when we start talking about data. And uh, I'll let Brandon go into a little bit more. But just in the hypersonic realm when you think about it, I mean the Navy's doing hypersonic testing, both offensive and defensive, uh, the army is doing hypersonic testing. We have the Air Force doing hypersonic testing. We have various uh, agencies that uh, like the Missile Defense Agency that are um, obviously have big buy ins to the hypersonic uh, defense uh, area. You have a multitude of um, commercial developers that are doing their own testing because they're trying to deliver capabilities uh, to the warfighter. You have uh, FFRDCs, you have universities, um, like UTSI is affiliated with the University of Tennessee. You have both air um, flight testing, uh, regimes that are going on on both coasts. And you have a multitude of ground stations that are doing ground testings and hypersonic wind tunnels to really understand this regime. And so when you think about it in terms of creating data, we can create data, we can put money at this, build things and create data. The key though is the next steps, which is how does one side of an organizational stovepipe understand what the other side is doing? And there may be a critical piece of information that uh, exists across that they don't necessarily even have any visibility on that they need in order to mature their models, in order to close their design. And so that really is kind of what spurred my interest. When I saw the reconciliation numbers come out and knew where this thing was going to go. It really comes down to okay, that's great, but how, how are we going to do this? What do we need to do technically in order to make sure that we can share data. And then how do we, what do we need to do organizationally? How do we break down stovepipes? How do we get industry to talk to research, uh, to talk to government and, and to pass along all these things so we can collectively move and rapidly move our understanding of the hypersonic, uh, flight environment forward. Because really that's one of our largest threats that we're dealing with right now. So you know, the problem wasn't, isn't generating data, we can throw money at that. The real issue is like the, so what, um, the way, you know, when I brought Brandon onto my team, you know, the way I described it is in the past, you know, there was a needle in a haystack and we were, we'd have to go find that needle. Now there's a needle in thousands of haystacks. And so you either need to like exponentially ramp up the human in the loop analysis of all this data or you need to go down this other path. Right, which is, uh, applying technologies to that. And you know, in, in order to do that there's a lot of front end work on that data that's required. And then you know, some of the things that, when we talk about, and I'm going to hand it over to Brandon is like, you know, data standards, data quality, um, data organization, like how are you going to meta tag this in a way and so that you can find it? And then I think one of the most interesting things is right when we talk about data sharing, you got to worry about data protection. And um, and so like how do you share data and protect it all at the same time? So to me that's really what spurred my interest, uh, in this topic and kind of facilitated at mtsi, bringing a team together to try to really tackle these big problems. So Brandon, what are your thoughts on that?

Brian Fox: Yeah, Brandon, especially when it comes to finding the needle among multiple haystacks. I love that analogy.

Brandon Frisco: Um, yeah, and to Gordon's point, uh, that's been a concern. I've seen that in every phase of testing, training and you know, building out tactics. And that comes down to data observability, uh, exactly what Gordon's talking about. And that requires tooling and like he said, pre processes. The biggest goal is everyone has this pilot data and they're focused on what they're looking at, the output, but they're not focused on being able to track it long term, being able to really understand it and what they have. Uh, if you add observability and that goes into, uh, eventually we can get to metadata and the tooling to create those, uh, those interesting, uh, artifacts. But that is really what gives you the durability. It lets, it lets you explain what you actually have, uh, what you currently need, what's missing from your. From uh, your current data. And then working with everyone to really kind of get a bigger picture. Uh, just like everybody wants to get the theater picture, you know, in combat, uh, you kind of want to get the same picture with all the data that's going on and that metadata is. And those pre. Processes that Gordon. Exactly. You know, hit the nail on is. Is really step into even starting that and, and reaching that pro. That goal.

Brian Fox: Absolutely. Ah, can you all lean into that a little bit? Um, because, Brandon, I love the way you described it as, as well as how you described it, Gordon, that like understanding what you might have, what that entity might have. But then there's also the need to understand what other entities might have, as you've already kind of described. There's. There's multiple government entities, there's multiple academic entities leaning in on this as well as industry. And, and there's going to be points of friction where, uh. And I hope no one in the audience minds me bringing this up, like, oh, Air Force, how well do they work with the Navy and then Vice, you know, utsi, how well do we work with. With Purdue or the University of Minnesota? And then meanwhile, you look at industry and IP is how they're going to survive. Right. And how much of their IP is tied into that test data. And yet for the national interest, regardless of what the friction points might be, understanding all the possible data that could be shared for the national interest, do you all mind leaning into that a little bit as far as understanding that big. Brandon, to your point, kind of the big theater of test and evaluation data.

Gordon Fogg: Yeah, I'll start and then I'll hand it over to, uh, Brandon. Um, like I said, there's multiple challenges of this, and I think that when you look at the Pete Hegseth memos, um, in terms of where he wants to go in implementing AI, everybody kind of focuses on chat, GTP and the AI agents and the tools and the end game. Um, but really none of that. I mean, as I was doing a lot of research on this last year, I saw that just in the commercial world, 80% of all the AI tools fail. And it's not because they're not coded correctly. It's because the upfront work of getting the data organized and how it should be isn't done. And so I think there's a lot of headwinds that the Department of War is fighting against right now. Some, um, of these are just people like myself that grew up in an analog world doing PowerPoint engineering and PowerPoint assessments and developing very, um, uh, boutique, um, like MATLAB scripts to look at, um, you know, to do test analysis. You know, that's uh, that's one, certainly one piece of it. You have, you know, acquisition limitations via contracts. We have the siege rolls that are set up a certain way that kind of support this legacy, uh, development. Um, when we talk about data rights, I mean, you mentioned ip, that's, you know, certainly going to be a huge thing. We, we have, we know for a fact that while the Department of War wants to avoid vendor lock, the vendors would love to have vendor lock. Right. I mean, because it locks them into that, you know, into that delivery, into that, um, capability for the life cycle.

Brian Fox: It's a natural point of friction there. Right. And the interests aren't, aren't aligned.

Gordon Fogg: Yeah. So it has to be attacked in just multiple, uh, ways in terms of like this data sharing and data information. And you know, one of the things that we started talking about on this was, you know, like the Library of Congress, I'm, like I said, I'm old school. I talk about, you know, the Dewey Decimal system of hypersonic test data. I mean, imagine that if there was no organization and you were supposed to go find a very specific book in a library, you would just be spending all this time going through it. And so, you know, um, we have an IRAD going on at MTSI right now of looking at, like, can we develop a mosa, uh, meta tagging standard so that at least there is a common methodology. So when you're collecting petabytes of data coming out of a wind tunnel, kind of series of testing or flight testing, that data that's tied to vibration, you know, on one side of the isolator of an IMU would be tagged in a certain way. And so we, so you start out with this kind of data structure and Brandon is probably over there cringing at, um, you know, my big hands, big map approach to this. And then, you know, the next thing you need to do is like, how are you going to organize this data so it's accessible, right. If I, if I take my data and I say I need to protect this data and I stove pipe it into a, you know, an on prem kind of server collection, and all I'm doing is analyzing and purging, right. I Collected my data, I put it on there, my analysis team looks at it, we know what we're looking for. We find that data, we, we spin that back into our design and then we flush the server in order to collect our next piece. Well, what did we flush? We may have flushed something that's very important to somebody else. And so there's this whole kind of like, test ecosystem across the various, um, stakeholders that, that needs to be figured out. And then we need to break down the stove pipes, and then we need to implement the things that Brandon is really, um, passionate about, which is like, how do we then develop like an on demand AI system like Netflix. You log in, right? It tells you exactly what. If you've been watching Warflix, it's like, here's your next things that you want to watch. Right? Um, and so developing a system like that, that is not just searching something magical, but it's actually telling you, hey, you know, that piece of information that you need, it exists and it's in, it's over there. So, you know, I think when we start thinking about big hands, big map on data, we need to attack each one of those, you know, constraints if we're truly going to get to a point where we can handle all the data that we plan on generating. So that's kind of my thoughts on that.

Brian Fox: Yeah. And Gordon, let me just respond to that a bit, because when I was at 18F and we were on a few projects related to AI ML, uh, one of the big takeaways, I'll use another analogy, moving away from the haystack is the need for a government entity, or an entity, um, to clean its garage before it got into an interesting home or car project. Like, the organization may want to start working on the interesting hot rod, but the garage is a disaster. Right. And that metadata tagging, all of that data work is uninteresting. It's like cleaning the garage. Are you ready to do this work? And yet you also bring up the Netflix example, but that's one stovepipe of data. They have to just manage their holding. Right? Um, whereas across government, across academia and industry, you're talking about data, mesh, et cetera. So, Brandon, with that, add to what Gordon was sharing and any other thoughts you have as far as enabling discovery and analysis across multiple past or stove pipes.

Brandon Frisco: Yeah, yeah, no, it's exciting to hear because that is, I mean, a perfect analogy. Uh, as Gordon mentioned, we all saw kind of chatgpt get really, really popular, right? And all of a sudden it got really improved. And then you know, uh, we thought no one would catch up. And all of a sudden, you know, uh, an internal memo goes in to Google and Google said, calls red alert. And then we're like okay, yeah, three months later everybody's like wow. Gemini 2, Gemini 2.5, Gemini 3. Like how did they do it? Well, Google is the, you know, the epitome of data organization. Like that's what they do. They just collect organizing stuff and put it together and you have this instantaneous shirt. Uh, search. Excuse me. Uh, so Gordon really hit there of uh, of uh, if you, that is the predefining system, like the setup to get AI to work. So he is 100% correct on that, on that. You know, all of those AI systems that fail because they, they, you know, 80% of them all fails because they assume AI can just be applied to a mess. Like you know, just, just throw a pile of data there, throw a, on it, it'll figure it out. But that requires searchability and, and time and it's you uh, know, information distribution by, by uh, by category. And if you don't have that, AI really takes forever to do its search. So that is the foundation. Uh, and to get that moving, uh, requires things like adapters, ah, and common exchange formats. Those uh, companies, they thrive on those types of systems. There's a reason why most of the um, the top seven companies, the Magnificent seven, I think they're called Magnificent ten now, they just keep growing, use uh, the Apache products, uh, because the Apache products themselves, like they leverage a lot of these universal tools that they can all talk to and interchange with. And so going back to what Gordon mentioned is interoperability is going to be important. Uh, so to do that is first get that standard that Gordon's looking for. Uh, and the government's kind of looking for as a whole, they've always been asking for it, right? They've been where can I buy Google? Where can I buy what they do? But that's tailored and they all, you know, do that individually and if they sold it just like the friction, right, they'd be out of business. If you could search like Google, why would you need, you know, services like Google to come in and do that for you? Uh, right, yeah.

Brian Fox: That is the goose that's laying the golden egg for them, right?

Brandon Frisco: Yeah, yeah, yeah. And they have that friction point, right? They want to be proprietary. They don't want you to have a government open source version. So it's going to be on the government themselves to make a GOTS product that really kind of encompasses that and start putting out those standards, putting out those requirements to start saying, let's have adapters for the data. We're not going to change the systems. They're redundant and they're usable for a reason because, uh, they have to be in play and they've been in play for years. So what we need to do is we need to adapt the data and get common exchange formats and move forward with a kind of a central ecosystem or a central kind of ideology to start talking together. I'm not expecting everyone to go into all the old hardware and start converting all the old hardware, but we can make endpoints that really start bringing that all together and really start kind of, uh, I guess emerging. An emergent property of that would be this ability to then speak to each other.

Brian Fox: Yeah, interesting. And I've always leaned on the garage hot rod analogy when it comes to data data preparation in order to start with AI. But as you all are talking about this, it's not even the garage. It's almost like being in a neighborhood and me relying on my neighbors and how their garages are organized, including my own, so I can work on my hot rod. Right. And, and that interoperability, uh, and this is within various government agencies, so you know, within Department of War, in darpa, ah, iarpa, anybody who's leaning in on this and running tests, as well as industry as well as academia. So it's a lot of different data garages.

Brandon Frisco: Um, yeah.

Brian Fox: And aside from that forward, like, what are some of the gaps there, whether it's data standards, policy, technology, branding. You brought up Apache. That's interesting. But are there other things there as far as policy? And Gordon, you brought up. Yep. Hegseth, uh, is, you know, kind of a big, big hand wave wanting to do this, but there's probably below that policy and regs that need to, need, uh, to be pushed or implemented. Any thoughts that way as far as helping move out on this?

Gordon Fogg: Now we're seeing this, you know, at the corporate level. We're engaged heavily across, you know, the Department of War through digital transformation. Right. And you know, and what we're finding is that, you know, we have a legacy we'll call acquisition workforce. And you know, they, they understand kind of at the very, probably at the very top level that where they want to go like this is what we need to do. We need, okay, we need to speed things up. And this is, you know, potentially how that can happen. But the implementation of truly the devil's in the details to get you to a point where you can get to like, you know, corporately, like what Google is doing when we think about, from an enterprise level. And you've touched on these things, right? Um, there was a report that came out, um, out of OSD when they were, they were looking at um, digital transformation writ large. And I, I see this as a digital trans, as a, you know, we talk about test, um, data and test data sharing and implementing. It's just a subset of a, of the digital transformation of how we can implement technology to wade through more and more data. And um, we talked about a little bit before we started recording that in the 90s data was kind of a dirty word. It was all about intelligence and information, which meant we need to process data so we can get to something actionable. And now with technology it's really the other way around. More data the better. And then we'll let you know, we'll, we'll let the technology sort it out. We'll compress that OODA loop even with more data through technology. But you know, the bottom line is across all our customer space, the implementation is really some of the challenging things. And you talked about for instance like policy. Okay, so um, you know, how are we going to, who's going to have the authorities to allow a, um, an FFRDC or a corporate entity that's running a wind tunnel, um, in a specific location to tie in to maybe a TRMC database, uh, to share their hypersonic data. So you know, you need these, you know, these um, you know, the proper authorities for connections and so forth to ensure like, that we're not putting bad data across those lines, that we're not developing lines that somebody can exploit. Right? So like sharing and protection, they're. There's a lot of things that are intention, right? Sharing an ip, those are intention. And it really is going to have to be driven by both technology. Like there's a solution that we need to drive to, not an idea, but in a solution that we need to drive to. And then, and then we need to decompose that solution, um, across all those means, technology policy, authorities. And then it needs to flow down contractually and how we are requesting and doing data rights and so forth. So there is a lot of heavy lifting that needs to be done. And you know what I think is really interesting is like two years ago nobody was really talking about data. We were starting to talk about aiml, but people were throwing those buzzwords around, like what did that really mean? And then we started getting more involved with aiml and now just Like Brandon was talking about, everybody's realized, Michael, we can't implement that without all this other front end work. And some of it's technically um, oriented or data organization, meta tagging, data quality, processes, uh, those kind of things, data sharing. Some of it is, like I said, policy. How are we going to share things contractual? How are we going to get, you know, how are we going to make sure that the people that we're asking to do this work, it's contractually anchored so they're delivering what we, what we need. Um, and so yeah, I wish I could say that, you know, that we, Brandon and I, have some magical, you know, a path to get to that. I think generally it's understood all these challenges, but how we attack we corporately in the Department of War, um, you know, and the partnership across uh, industry, across the service contractors and across government itself, they really need to come together and figure out how there needs to be like a massive data summit and with a plan to, to drive down how this is going to happen. Without that, I think what you're going to get is people are going to attack it in a piecemeal fashion and you're going to end up with what I call Legos and Lincoln Logs. You're just going to have, you know, disparate solutions that don't bolt together.

Brian Fox: Yeah. And this topic has been really interesting to me because as a geographer, uh, there was a challenge as far as data discovery, um, where as a mapper, uh, whether I was in the intel community or at a science agency, uh, the biggest amount of work was always finding and then downloading the data. The analysis was relatively easy, um, but that's gotten harder and harder because there's more and more data sources. Meanwhile there's some open source capabilities like Stack and Parquet or geopark that are connecting disparate data sources together. Um, and then that is enabling AI. So Brandon, for you are there because Gordon just described maybe it's an industry partner or ah, an academic institution that dumps data into a central repo on the geospatial side. That never happened. I thought it might with, with the standup of the cloud like 10, 15 years ago and the government moving towards that and it's actually been data mesh like uh, the idea of tying that together through open source capabilities like Stack and Parquet. Geoparke, are there some technologies that are out there to help tie together what already exists and just allow that discovery and if it's properly tagged, as Gordon was leaning in on, um, some analysis, any Thoughts there on the, on the tech side to enabling discovery?

Brandon Frisco: Uh yeah, no that's actually the current paradigm in industry is not to move your data if you don't have to. Right. Data. Moving data is costly and Gordon uh, likes to speak a lot about federated data which is very important. And that's true. You can have your data in different locations and that's fine. Uh, these tools leverage that. They also have policy built in, in the front as long as you have the policy. And to that point I agree, I've been trying to kind of advocate for uh, policy needs to be um, kind of generalized from top down so everyone kind of follows the same uniformity. It's not thou shall um, thou shall just make metadata, it's let's follow this format or not to that specific, I don't want to be too into everybody's uh, business but it's basically here's the paradigm you should follow. So they are all kind of universal um, to that point. Uh, uh, a famous horse trainer, uh, my sister loves horses, uh, Pat Pirelli, he says take the time it takes. So it takes less time. Right. And that's really what these tools do. If you get these tools in there and you use something like geoparkay, I mean you hit it right there. Parquet Avro metadata, uh, systems like open Metadata are used. They're all built for large scale disparate uh, data across geo locations. And if you take the time to set these networks up and actually locate it and get it all integrated then you're going to have that observability. That's exactly where that comes from. That's how these companies even operate. They wouldn't be able to operate globally without this, this you know, this ability to, to be able to look and search across everything they have and own and then validate it. And uh, so some of these tools are things like open Metadata. Netflix uh, themselves created a large scale tool they needed to use called MetaCat. It's open source, free to use. You can just you know, get it off the shelf basically uh, and start tying it into your own purpose. So so you know, kind of wrapping that all up. Policy really dictates the direction to go. Um, then these tools, as long as you have the right direction, really facilitate that and will get you into a place where they start to fall in, it just starts to fall into the right, the right format. Uh, as long as you're, as you're following the current path. As you were saying, as long as everybody's starting to clean their Garage first, like, you're going to start to find out a best way to organize things, and then you're going to kind of step back a bit and look at it and be like, you know what? I could do this a little better. I could do this a little better because everybody's on the same page. And then all of a sudden you're like, hey, now we can all kind of understand what all our neighbors are doing.

Brian Fox: That's all fascinating. Where do you all want to go next? Uh, in this conversation, like, what have we not hit upon? We've talked about some of the things that need to be done. The reason why. What else am I not bringing up that you want to share, Brandon or Gordon?

Gordon Fogg: Well, I think the biggest thing is that the solutions have to be driven from a hierarchical approach down. I think that in terms of, especially when we start looking at test data, um, um, you know, in the hypersonic region or the missile defense region or any of those kinds of things, because, you know, companies are going to ask, act in their best interest, organizations are going to try to move forward in their best interest. And that's great until you try to integrate things together. And, you know, one of the capabilities, you know, uh, that we have at mtsi, we do a lot of modeling, simulation, uh, work. And what we've found is, uh, when you get these different models, some of them are, you know, are built inherent or organically within mtsi. A lot of them are coming from, you know, different organizations. And when you try to, you know, stitch it together to understand, like a kill chain from end to end, whether it's a blue or red kill chain, you have these, you know, these models that don't necessarily talk to each other. And then, so you're, you're spending all this time like, building translators, uh, to get these things to actually be able to work together across, you know, in an entire scenario. And so the, you know, I think the big thing is, you know, that organizationally, you know, the rush to a tool, I think, is, uh, is a concern that I personally have. I see it in my customer space of where it's like, you're thinking very tactically, like, I need a tool that can do something right now to help me get this thing to CDR or pdr, whatever it may be, right? And you go out and you find this tool and then it has to go through approvals to, you know, in order to get loaded up onto the system. And then, you know, you're now managing licenses and all of a sudden it becomes irrelevant or it becomes ob by you know, um, design maturity. And so you're now chasing this next tool without really thinking of like where are you going to end up, you know, where is the end state of this? And so um, you know I believe that like uh, very much of having a strategy and somewhat of a top down approach on how you're going to get there. Um, you know I'm a strategic planner by, you know, by my Navy experience and you know the tactical actions should all be traceable up to an end state, up to an operational plan in a strategic end state. And I think right now what we have is a vision of where we want to go, but I'm not sure we have a strategy. Um, and I. So you know, to me that's what I'm keenly interested in and helping you know, shape my customer space to get out of solving next week's problem and start thinking about where do you want to be two years from now? Because that's probably what it's going to take to implement all the things that we discussed, whether it's technological innovation, setting up data structures, getting the policies in place, amending contracts so that you can support all that. So that's kind of my thoughts Brian.

Brian Fox: Yeah, I appreciate that. On the tools, you know, um, I got to see that in the intel community uh, on the geospatial side where um, at times there wasn't enough vendor lock. Right. Like, and in other cases it was too much. Uh, but uh, I don't know, equating it back to like a home project, it can be hard if there's a variety of tools necessary from a Allen key to a Torx wrench to Phillips uh, head. And it's random, right? And that's kind of what you're describing there that folks are making. I'm going to use a Phillips head on this part of the project and over here someone else is deciding Torx, a Torx head and standing back you're like now we need 15 different tools and none of these things uh, tie together. Um, is there a need for policy decisions and maybe open source opportunities, uh, the government to invest in open source. Um, I know that helped a bit, uh, especially out of in Q Tel pushing for open um, source on the geospatial side, um, to stir the pot a bit. What are your thoughts that way as far as uh, the tools that are available to enable the analysis of test data and uh, what more could be done?

Gordon Fogg: Well I'm going to go to Brandon on that in terms of tools he's just a lot more knowledgeable on, you know, on data structure tools and so forth. I think on the test analysis side, um, you know, looking at those tools and making sure that you're not putting a square peg into a round hole kind of thing. Um, and, you know, ultimately the, you know, doing the analysis on the data, we're very good at that. Right. The problem is, is that when you dump so much more data in there than how do you know you're looking at the correct data? How do you know that it's quality data, how all those kind of things, you know, analyzing bad data is, uh, not good. Right. So, you know, in terms of. I think it all comes down to once you get the front end, um, in a fashion that people can find data that they have confidence in, then the application of the analysis piece, in my opinion, is a lot more simplistic. But I don't know. Brandon, what do you. What do you think?

Brandon Frisco: Yeah, yeah. I mean, to that I have really been focusing on exposure and to that point where the models and the tooling matters, as Gordon kind of hit right there, because I've been trying to say that very much. Uh, I've been trying to say that a lot. Excuse me. Uh, is that really digging into the reason for the tool and why it exists is kind of important. And that comes really from exposure. So my, my whole passion of this has been bringing education and exposure and observability to how industry has done it. One, um, kind of concern, uh, is we deal with computers all day at home, and we have this kind of understanding of that's what a computer is, this is how it operates. But then we kind of disconnect the need to think of scale, to think of the future path as kind of Gordon says, what do you want to do six months down the line? And when you look at scale, you can't build something the exact same way. You have to. You have to build it a little differently. You can't use, uh, for instance, like in water, you can't just use one pump, um, then just make it ten times larger. You have to think of physics and then you have to kind of separate them and end up using 10 pumps to do the same thing, uh, and combine them and find a way to do that into a manifold. So as that example, uh, you know, exposure to really how these tools are used, how they're used in industry. There's a lot of knowledge out there. There are tech blogs that, uh, Netflix, Instacart, uh, Uber. They all explain how they've solved these problems and they explain how they've used tools at scale. A new one just came up by OpenAI. How they, uh, scaled up post res, which is a, you know, uh, it's been a database around there for, for years now. Uh, decades. Yeah. And they just showed how they were able to scale it and paralyze it to service 800 million customers. 800 million queries per second. And they explain the whole rationale of how it's done. They don't give you the exact details, obviously, that's their kind of secret sauce. But you can look at their observabilities and then copy their ideas and apply to your own problems. And so circling back, my thought process is if you get the policy in and then you start sending out kind of these signals of take a look at this, observe this, work with each other. Stop just applying the tool. Like you said, uh, where I'm using a Phillips, I want to use a torque wrench. Why? Like why? What problem are we trying to solve? Who has tried to solve that similar problem? And then kind of massage it and integrate it into what you're doing? Uh, so my key points are really, especially, uh, in the academic field is start providing exposure, start providing education. So that way everyone who's. And I get it, I understand you're busy at your work. You know, you're focused on that. You're doing that all day. And where's the time to really, to really stop for a second and kind of read some stuff that's going on in the world. And, uh, that is difficult. I find myself struggling to do that a lot too. So, you know, kind of stepping back again, taking the time it takes to take less. So it takes less time is kind of important. And that will facilitate exactly what we've been talking about this whole conversation. It will really start. Everyone will start clicking. They'll start working together because they'll see these things and they'll think of it in a different way and they'll say, wow, I didn't think to use that tool this way. I didn't think that we could actually not even need three tools or 12 tools. Maybe we do need some vendor lock here because this really leverages that scale we need. And this over here we don't. So we can find the right middle ground. Uh, what you're explaining.

Brian Fox: Absolutely. Really appreciate that for both of you. Is there, we're coming up on the end of the podcast here. Is there anything I didn't ask that you'd like to share with the audience and Gordon, I'll start with you, sir. Is there anything I didn't.

Gordon Fogg: I think the other, you know, we talk about data sharing and so forth and one of the other challenges is just, you know, we talked a little bit about like IP versus open source and so forth, but we, the other challenge we have really is, you know, the levels of, of data that's out there, right where you potentially are capturing hypersonic data that may be just CUI or on class. And then, you know, some other data may, you know, indicate, once you start processing it, it may, you know, indicate a vulnerability. And so all of a sudden it goes up to higher levels and um, you know, some, you know, there's, there's work in various very high channels, whether it's TSSCI or SAP. And then, you know, the ability to kind of like, uh, figure out how you can work the, the data up and down, how you can take maybe the key pieces that are at this very high level and distill out the CUI piece that somebody else may need at the CUI level. And, and that's both. And when we talk about like the totality of things, I think that's a microcosm of the, of like the big issue, right, which is sharing versus protection. It's about technology. How do you, how do you take things that are at a certain IL level and you know, and move it down or move it up and so that people, you know, have access to it. How do you work those kind of connections? And that's a policy in authorities, um, how do you protect it? And so I think, you know, there's, there's a variety of challenges, but you know, guess what? The world keeps moving. Everything's, you know, the challenges are why engineers exist. That's why, you know, we enjoy coming to work every day is, is trying to work through these and you know, and trying to develop solutions and technology spinning very quickly. We're probably find solutions to problems now and you know, at the same time new, you know, challenges will emerge as new technologies emerge. And so it's, it's certainly an interesting time, um, in terms of what we're trying to do. And I think the vision is, is amazing and great. You know, I think it's now just kind of getting everybody on the similar vector, uh, so that we can work through these challenges and not, you know, duplicate effort, uh, to be efficient and effective across Dow and uh, share you know, um, lessons learned and things that have worked and not work, reduce some of our risk intolerance, be more risk tolerant. To trying to drive quick solutions. Like I said, all these things are, we're going to have to change as we migrate from a kind of a legacy acquisition approach to a new digital transformed uh, approach. And it's exciting to be part of this and I really appreciate the opportunity to give our ideas, uh, certainly my ideas and how we're thinking through this.

Brian Fox: Oh absolutely. And I think this will help stir the pot a bit and drive more discussions around it which hopefully helps move the needle even further. Yeah. Brandon, anything I didn't cover or ask that that you want to share with the audience?

Brandon Frisco: Uh, yeah, I mean I really think just taking the first step and actually you know, looking at these and trying to solve these problems with these, the concepts in mind, uh, will be kind of the big push because uh, the hardest part is starting. Right? Um, and when you start and you end up in discovery and to mention what Gordon said, a couple of things like the data tag in the organization and things like that, those will eventually fall out of starting to use these types of ideologies and these paradigms because you'll inevitably start figuring out how to build those into the policy based tools and everything else that evolves. So, so you know, starting that and moving forward and you know, taking the initial steps to say okay, is there a way to do this differently? Can we do this as the Department of War as a whole entity and really be more uh, like, like he said, less uh, risk averse and saying you know, okay, there's going to be some issues but we'll figure them out. We have a way to do that in the future. We're not going to be stuck there. There's an ability to pivot and adapt and that's, I mean that's, that's what everyone's good at. That's what we're good at. That's what the department does. It adapts. Right. And you know, with that mindset moving forward then we can accomplish those, those ideas. So again, I mean I, I thank you for this and, and at least uh, uh, an ability to a ah, medium to, to kind of give everybody some kind of insight so we can kind of spread this, this information and hopefully just start, you know, getting everybody to, to at least kind of think for a second and say yeah, you know what, actually I'll listen to that and, and maybe somebody here and there takes a look at it and finds an amazing solution that then we all get the benefit of.

Brian Fox: Agreed. And at some point in the future I would love to have you all back on as Part of a virtual panel. Maybe that's something we can plan and work towards in the coming months, is getting a number of folks together from a variety of entities. You know, you all are coming in from industry, but maybe some folks from government, uh, some folks from academia and stirring the potential, continuing the conversation. But, you know, getting the snowball, building and moving forward, that would be so much fun. But Gordon and Brandon, thank you both so much for being on today and everyone that's, uh, listening in. Thanks for joining us today. I. I hope you enjoyed today's podcast. Uh, it has been so fun pivoting this towards space and defense. So this is really always interesting and, and fun to lean in on on this area. We're always interested in hearing from technologists and technology leaders in government, academia, and industry. So if you'd like to share how you're driving innovation and modernizing outcomes for the U.S. space mission, please reach out to me at bfox24tsi.edu. Again, that's bfox24tsI.edu. And Gordon and Brandon, thank you again for being on today. Really appreciate it.

Gordon Fogg: Thanks, Brian. Um, appreciate the opportunity to have this discussion and look forward to any future discussions you want to have. Joining a panel sounds amazing.

Brandon Frisco: Love to do it.

Brian Fox: Yeah, that would be a blast. Yeah. Thank you both. Really appreciate it.

Brandon Frisco: Thank you.

Brian Fox: Thank you all for listening to today's episode of the Space and Defense Innovation Launchpad podcast. Don't hesitate to reach out if you'd like to be a part of a future podcast. We'd love to hear from you. We hope you enjoyed today's discussion, so please don't forget to, like, follow and subscribe so you don't miss out on, um, future episodes. We'll see you next time on the Space and Defense Innovation Launchpad podcast.