The Flux by Epistemix
Welcome to The Flux - where we talk data, decisions, and stories of people asking the what-if questions to create an intentional impact on the future.
The Flux by Epistemix
Revolutionizing Engineering: The Power of AI and Simulation with Dave Freed
In this episode of The Flux, join Dave Freed, Senior Director at Ansys, as he delves into the transformative world of computer simulations. With a rich history from Exa Corporation to OnScale and Ansys, Dave explores the evolution and future of simulations, highlighting their pivotal role in automotive, aerospace, and nuclear engineering. Discover how simulations have advanced to replace costly physical testing, improve accuracy, and prevent errors. Learn about intriguing real-world applications, the integration of AI in automating complex tasks, and the surge of simulation in life sciences. Dave also shares personal experiences and acknowledges mentors who've shaped his illustrious career.
00:00 Introduction to The Flux Podcast
00:19 Meet Dave Freed: CTO and Simulation Expert
00:41 Dave's Career Journey in Simulation
11:59 The Value of Simulation
22:18 Simulation Accessibility and User Experience
31:49 Future of Simulation and AI Integration
Hey there. Welcome to the next episode of The Flux.
Today, we're on with Dave Freed, who's been the CTO of multiple simulation startup companies that have been acquired. It has been through the largest acquisition of a simulation company in history when Ansys was recently acquired. So [00:01:00] Dave, why don't you give our listeners a little bit about you and some of the roles that you've had in the past.
Dave: Yeah, sure. Thanks, John. Great to be here. I'm currently a senior director. In the, cloud AI solutions and enablement business unit at Ansys corporation, which, is going through a merger with a synopsis. As many folks, are probably aware, it's been in the news a lot, but I started my career, My professional career at Exa corporation, a small startup that was spun out of MIT.
My PhD advisor, Kim Mulvig was the founder of Exa and Exa was completely focused on a revolutionary way to do. Computational fluid dynamics, a particular type of simulation, but definitely one of the most common, application areas for computational simulation, what people sometimes call engineering simulation or physics based simulation.
And, I was at XF for [00:02:00] quite a long time. I had a little, I call it my sabbatical where I went into, data mining and machine learning,back before it was. the biggest thing on the planet and eating the world. This was back in the early two thousands. but, but yeah, most of my career, has been in simulation.
So I was at X for quite a while. And eventually we got acquired by Deso system where I continued, in a management managerial role, overseeing, a number of different.industrial areas, including oil and gas, various kinds of multi phase flow applications with, with, application to, environment as well, ironically, along with oil and gas.
And, and a few other things. And then I left to,in order to join a startup called OnScale that was completely focused on making simulation work well on the cloud and making simulation much more broadly [00:03:00] accessible to small to medium sized companies and even individuals, academics, by offering the resources available on the cloud.
But in an easy to get started. Way that was cost effective and straightforward to jump in and start doing things with simulation. And we've got a little ways down the road with, the on scale mission when on scale was acquired by Ansys, in order to become the cloud platform for the, large number of Simulation applications and products, developed by Ansys, which is the world's, leader, the market leader in physics based, simulations.
So that's been my journey so far and, really enjoy it. I'm very passionate about physics based simulation and, and making it work well, in new ways like cloud with GPUs and that sort of thing, and now, the role that AI is going to play [00:04:00] in, In simulation.
John: So Dave, one of the things we like to ask everyone who jumps on the podcast, how did you get into simulation in the first place? What was your kind of aha moment and what'd you do next?
Dave: Well, that goes way back to my, my time as a graduate student at M. I. T. I was in the nuclear engineering department, and within nuclear engineering, I had decided to specialize in nuclear waste disposal. Seemed like a pretty interesting, complex, challenging topic. and that turned out to be true.
It was a very interdisciplinary topic because you had to understand like some politics and economics and human factors and all this stuff. and at the time the leading proposal in the world for nuclear waste disposal was, to bury it underground in what was called a nuclear waste repository.
we would dig a deep tunnel and, create a place [00:05:00] to put the waste and bury it there and try to monitor it and,make it so that, the radioactivity, the radioactive materials would be safely, removed from the environment and people. And the concern, the biggest concern with this approach was the idea that over time groundwater would contact the waste form and dissolve the waste and carry the radioactive materials underground into aquifers and other places where it could reach the biosphere.
And interact with people, animals, you know,end up in places where we don't want it and to really get at this issue required modeling the flow of groundwater and that required a computer simulation, at least, to be sophisticated about it and to be as predictive as possible. And so that really got me into the whole idea of computer modeling and simulation [00:06:00] and.
I became really, really obsessed and entranced with the idea that by formulating the rules and the algorithms properly, we can use a computer to predict and recreate. What will happen in the future for some physical process or system and we have to, make some assumptions and get those assumption assumptions correct and take a lot of things into account.
But if we do a good job, we can have meaningly predictive simulations. with some error range, of course, attached to it, you're not going to get extremely precise answers. Most of the time, too, if you're asking a very complex question, and, what, how much radioactive waste will be carried by groundwater,to the biosphere.
Over the next million years is a complex question. but that's started my, preoccupation with this whole idea of computer simulation. And from there it grew to flow [00:07:00] in general, which is the field of computational fluid dynamics.and then lots more things from there.
John: over your career, you've been involved with a couple,Acquisitions of these simulation companies. where have you seen like the growing demand for simulation come to life the most across industries? Or do you think this is just a natural progression of going from stats based interaction with computers to then machine learning to then simulation and now AI?
And then, simulation going to be growing because of more people getting into computer science or the simulation had its heyday. And now AI is going to take over or what, where do you feel like simulations going?
Dave: Oh, wow. Great question. I think that simulation and AI will be combined in ways that will cause the use and application of simulation to just explode. I think we've barely scratched the surface of. What's possible and the value to industry that is [00:08:00] possible with computer simulation.I think today simulation is maybe, 30 billion a year, maybe 20 billion a year, revenue.
area, but the potential is vastly larger than that. and I think, the critical bottleneck has been the complexity of using simulation. So it's the accessibility, it really requires a deep expert. In fact,an entire team of deep experts to, really get the most out of it and provide value.
But I think,with the help of AI and,cloud and, GPUs and other technology, we're going to see that accessibility issue,get handled and dealt with in new creative ways so that, engineers and professionals all over the world. We'll be able to take advantage of simulation.
Without necessarily even knowing that they're doing that the simulations will run in the background. They'll run very quickly. And [00:09:00] now, instead of a medical clinician or a construction manager, or,an architect, architectural engineer,using a computer program that's running some kind of simple linear fit or some simple model.
It'll be running a whole suite of complex simulations in the background to deliver, a much more comprehensive result, that really, increases the efficiency and,accuracy that, professional, needs, And there'll be a lot of value in that. And so companies will pay for that.
And I think that's the future of simulation and AI will have a really important role to play. And we can talk more about the interplay that. That I think will happen there. just going going back to, some places where we saw simulation really take off and start to become important and add value.
maybe one of the Uh,first areas was in fact in nuclear engineering. there's a lot of very complicated [00:10:00] physical processes happening there. And in order to design and develop the nuclear power plants, it was necessary to, create world class computer simulations to augment what was even possible with physical testing.
And that's a, an important theme of simulation is,where can simulation either augment or replace the physical prototyping and physical testing that, engineers, designers, engineers, analysts would be doing if they didn't have computer simulation.and then from there, computer chips, right?
chip design very quickly became much more than you could do with pencil and paper or even a simple program. and so computer simulation of, chips. And various kinds of electronics, became, just, completely, Embedded within the commercial process within the industry.
And all chip design, today is done with the assistance of computer simulation.I think we saw [00:11:00] in the automotive industry or, and, transportation industry in general, the rise of aerodynamic simulation and also, crash. Using structural simulation. so you don't have to keep crashing a lot of cars or, especially if you look at something like a train or an airplane, you really don't want to be, intentionally building one and crashing it.
there's a, huge value, to the industry and being able to do digital, crash testing. And that's a very important, and very now I would say, pretty mature area of, physics based engineering simulation.and then, beyond aerodynamics and crash,we've seen the automotive industry, really play a leading role in utilizing simulation and growing the industrial applications and the business associated with simulation, automotive and aerospace both.
we've seen a lot of, simulation of various complex [00:12:00] aerodynamic phenomena that occur You know,with airplanes and flight taking off and landing, are quite complex and benefit tremendously from, computer simulation.yeah, I would say, that kind of, takes us through,the last several decades of where simulation has added a lot of value.
John: So Dave, some of the listeners on this podcast, have a ton of simulation experience. Others who are new to simulation might be using this podcast just to try to understand the value of using simulation over just classic statistics or some mathematical modeling. if you had to describe across those different verticals or different industries, what common values of simulation apply, how would you walk somebody through that who's new to running simulations in the first place?
Dave: Oh, good question. Well,the value of simulation is that it [00:13:00] will, really allow, much less constraining assumptions.if you're trying to use a relatively simple model, you'll have made some pretty big assumptions that model, that, that have to be true for that model to be accurate.
Whereas, these physics based simulations or even other kinds of simulation that Are more complex and require more compute power,the reason what you're getting for that extra compute power is the ability to attack problems from a more fundamental point of view. And so now you know, you.
If you can answer questions like,what are the materials and material properties of this device or system that I want to simulate? And like, how are these, how are the different pieces attached to each other,and what are the loads on this structure or,this flow problem, or, this electrical [00:14:00] system,you can now predict to within, I would say usually a pretty useful range of accuracy, the detailed physical behavior,and in a, in what's called a transient or time dependent simulation, you can even see how the device or system will behave over time.
In a much, much more detailed way than you would get out of,let's say an analytical model or a, typical statistical approach.
John: Dave, one of the things that, we try to understand across the different types of modeling that people get into, what are some common, beliefs in physics based modeling and simulation that you might disagree with or that have been overturned in the last couple of years as Either computing has gotten more powerful or models have become better over time.
Dave: I think the big question around physics based simulation and probably all simulation, agent based [00:15:00] and,statistical models, Bayesian approaches. there's.so many different, kinds actuallyis, reliability. And how do I really know, whether this prediction will be reliable?
And so there's, I think, always been the feeling that. we need to go back into the lab and really test this thing physically. We need, we still need to build the prototypes to verify and, so what good a simulation, but I think the more nuanced way to look at it is that,it's true that.
a wrong assumption could be made, a wrong, there could be a wrong input or even the basis of the simulation may have flaws for any particular problem or application. And you are going to want to do some degree of physical prototyping, but simulation can help drive you in the right direction.
It could help drastically reduce the amount of physical prototyping that's required to where you, you can have some balance where, you know, 90 percent of the engineering effort could be [00:16:00] digital and will be much more efficient. By being digital because you don't have to build everything and, test it physically, which is usually much more time consuming.
and, maybe that last 5 or 10%. is still done with physical prototype, but really as a final check. And that's how a lot of disciplines have evolved. And so I think,this question of, reliability and accuracy, was a big question for a long time and still remains, but it's been addressed more and more by people realizing how to use simulation in the right way in conjunction with traditional testing techniques.
And also by, the improvement of various improvements in the simulation, accuracy and quality. And also some improved science and techniques around uncertainty quantification and how to think about the uncertainties associated with a [00:17:00] computational simulation and modeling results. So all of those combined have led to a situation where simulation is more and more accepted as.
A correct answer and a reference answer. And you see now that, if there's a new kind of simulation, very often it's compared to an existing traditional form of simulation that's treated as a reference as opposed to having to do. experimental tests to form the reference because that traditional simulation tool for that application will be considered mature and will have been so well verified against experimental tests that it can be considered the reference.
And we, that has really helped with the industrial adoption and growth of simulation as a part of industry.
John: with simulation having more of the referenceability and reliability over time, you'd think that simulation would [00:18:00] continue to be used more and more. But,you've been working long enough where you've probably been privy to a number of decisions, or at least aware of different types of decisions where Simulation might not have been used and on our podcast, we like to talk about this as a what the flux kind of moment and so from your experience, if you could rewind the clock and go back to, you know what, we probably should run a simulation of this or,I think that, whether it was a political decision or something else, it would have been pretty nice to have a simulation of that before that played out.
To avoid some unintended consequences. Do you have any examples of what the flux moments from your career that you'd be willing to share, or even just speaking more broadly about, a scenario where simulation could have been applied to have prevented some unintended consequence?
Dave: yeah, I think I have a good one. That I can share from really quite a long time ago. This would have been, probably about 15 years ago.at XO, we developed, I think, a pretty unique market leading [00:19:00] capability around prediction of an aeroacoustic phenomena. So that's when you have a flow induced noise.
and, you and I think many listeners will have experienced the aeroacoustic phenomena of Buffeting, which is when you have a sunroof and you open the sunroof or you open a rear window and there is a horrific sound. That gives you a massive headache and you start yelling at whoever's in the backseat to roll up the window.
or if you have a little bit of experience with this, that if you crack the front window, it will drastically reduce the amplitude of that throbbing noise. the, jargon that we used for this phenomena, was a sunroof or side window buffeting. What it really is a helm holt's resonance, for anybody who wants to go look up helm Holt's resonators, your, the, passenger cabin of your vehicle is acting like a large helm Holt's resonator, and creating a, a resonance at a [00:20:00] frequency that tends to match one of the resonant frequencies of the human skull.
Which is why it's such an unpleasant experience. And, we were working with, one of our early adopters, one of the big three automakers, in Detroit. And, they had,they had released a version of an SUV that had a pretty, pretty gnarly, rear window buffeting issue. And, we. showed them, through simulation, how, what the root cause was and some things that, that they could have done with the design that would have, made a big difference in reducing the severity of the problem.
And they absolutely did not believe it. they didn't think we had the cause of the issue, correct. And, or that simulation could do any of this. And so we showed them, the results of some simulations where we had made these design changes. And, they,were very skeptical.
And so we challenged them to do a test drive in one of the SUVs where, they [00:21:00] had actually implemented a couple of the changes in order to demonstrate. And,I didn't get to go on that ride along, but one of my senior technical people,went to Detroit and did the experiment.
And the simulation was absolutely correct. And I, the, lead engineer who was involved was a big champion, after that, and was really, was really amazed and realized okay, if I had this simulation capability and had been using it, It would have made a big difference in our ability to, design, ways to avoid, this severe of a rear window buffeting problem.
John: but I'm guessing that probably had some impact on maybe adoption of that vehicle. some expensive changes that might've needed to have been made in future models. Yeah.
Dave: Absolutely. Absolutely. Yeah. It, one of the, one of the keys to reducing the severity, is adding a, the equivalent of damping,the amplitude of that resonance [00:22:00] can be impacted by damping. And it can also be impacted by,what is actually forcing the resonance, what is it that's actually like pushing on the spring.
And it turns out it's an aerodynamic phenomenon. It has to do with, the way that vortices are formed at the opening of the window, and there's some, small design changes that would have reduced the, let's say the, magnitude and efficiency of that,of the, of those vortices being able to force exactly the right resonance frequency, causing that phenomena.
That was a great, what the flux moment for us. And we had a big champion and did a lot of good business with that customer, in, in our acoustics, probably that's. That relationship is still going.
John: Nice. So I'm gonna go back to one of the things you mentioned earlier in the podcast and talking about sort of the accessibility of simulation, which sometimes comes down to education and training. And I know [00:23:00] from one of our earlier conversations, you'd mentioned that While you were helping build AXA up, there was another competitive company, similar space, and they also were trying to get adoption of simulation going more and more.
would you be open to sharing some of the stories between AXA and that other competitor and how they ended up pushing on education and making simulation accessible through education, versus the vertical specific approaches that your team took?
Dave: yeah, absolutely. I think, the, ability to just get going, with simulation is one thing. the other part of accessibility is,hiding a lot of the complexity behind, A user experience that is intuitive and speaks the language of that user, and it's not that the user isn't an expert, it's just that the user,probably isn't an expert or you don't want to require that they're an expert in numerical [00:24:00] simulation in computational simulation, but they could be an expert in structural analysis or, you know,some kind of, health clinician or, construction or architecture or whatever.
and so if you can hide the, complexity that simulation engineers face. Then you can make it much more accessible and, but that's hard work because then you have to use a lot of automation and a lot of kind of clever design to automatically do all the things that the simulation expert would have needed to do, or at least just make them easier and more accessible.
Straightforward, easier to understand. those are the two sort of keys to making things more accessible. as far as a competitor, I don't know if I really saw anyone yet. Do a great job of,of the second approach where you,really have a well designed user experience that hides a lot of the [00:25:00] complexity.
there's certainly lots of examples of that, but very often it tends to be like a methods developer at a company who. really just makes a nice, simple, easy to use application for internal usage, because they really understand what, their team or their company needs, and they want to enable analysts, designers, engineers of various flavors and stripes to be able to use this simulation based capability that they've developed, but they know they can't expect all of those people.
To be experts on it. So they create a nice template, of some kind that's really easy for those people to use. And so what I haven't seen yet is a company really doing that on a broad scale where they,identify dozens or even hundreds of applications that are targeted and have a really nice, really well designed.
user interface with the [00:26:00] right experience for a specific target market and then automate and hide a lot of the simulation complexity,behind that interface so that their end users don't actually have to worry about it. So that's a big part of what I think is needed to address that accessibility problem and something I'm really interested in.
I don't think that design is,Where I shine, I, I'd love to be a designer, but I'm not that great at it. but I work with a design team that is really talented and,I'm always, just advocating, put yourself in the shoes of the end user, try to think like your end user and make sure you're not burdening them with anything that we don't have to.
That they shouldn't have to know there are saying some things they should have to know, okay? If you are a structural analyst or a device designer of almost any kind of device, you do need to know what materials your [00:27:00] device is made of. You are responsible for that. Okay? We can't select the materials for you, there's a lot of things we don't need to burden you with.
That we should just take care of behind the scenes,what sort of computational mesh is needed and, various numerical parameters that are in, very often inputs to simulation tools today that require, lots of experience and expertise to determine the best settings, the best values for those settings.
let's hide all that stuff as much as we possibly can.
John: for those listeners thinking about simulation and user experiences, what you're an advocate for is making sure that the outputs are very tailored to that end user and the way that you're interacting with the simulated data is specific to those use cases and you're minimizing the things on the front end, which, in some areas might be like model construction and other [00:28:00] things that.
You'd hope somebody is paying attention to, there might be a minimum set of things, but the rest is masked until they're able to interact with the simulation results to then help inform the decision. Is that kind
Dave: Absolutely.
and even the results, I think we have to really pay attention to what is the burden that we're putting on the end user to extract value from the results. if it's too difficult and requires too much expertise, there's a big risk that value will never be extracted.
And so it's really a,again, it's a design problem to put yourself in the shoes of the target end user and figure out, okay, what's the flexibility and the, the things that, we do want them to have control over and to be able to adjust, but what are the things that Are just a burden, and we can do them ahead of time or abstract them [00:29:00] or automate them so that it's either removed or just, at least minimized for the end
John: So I know that there'll be some investors listening into this podcast and when they hear words like automation, they might be thinking, is that the role that AI can play is to. Help tease out insights from simulations more quickly is like, how soon to reality do you think that is in the physics based simulation world, knowing that, if it's happening there, it's probably also able to happen on the agent based simulation side as well.
Dave: I absolutely think this is an area where AI has a role to play, and I think that we are going to see AI. Generated value, especially around results interpretation very soon. I think that is near at hand, because it's just a great fit to, for example, ask, an AI system to examine an image, or a set of [00:30:00] images, or possibly even a video of computer simulation results.
And,make,observations about those results or even, ask the AI to compare the results from, and different designs or in different input values. And report on, here's what I observed. think about the time savings of that compared to asking an engineer or an analyst to go look at all those results and, zoom in and out and rotate and, adjust and, do that very carefully and then create.
systematic comparison across 10 different, let's say you're doing a design study and you have 10 different designs and you want to understand,and there's some, ranking of those designs.let's say you're trying to find, let's say it's a structural analysis and you want to figure out, what will meet the load.
the required load on some part of the device, with the minimum overall weight of the [00:31:00] device or something like that. And you have these different designs and you want to understand not only, you know,how they rank, but get some insight into,what. Is the physics behind how this, lower weight, device was able to, meet the load requirement.or, in fluid dynamics, you very often have very complicated phenomena going on, So the ability to ask an AI system,what's happening here and what led to this, reduced, Aerodynamic drag value and then just, quickly compare that across a large number of designs.
That's like almost tailor made for an A. I. M. L. System on even just being able to ask questions about, show it a results image.what do you see in this image or show it a plot, of, that could be anything, a set of S curves from an electronic simulation, and just ask, Hey, what do you see in these plots?
compare these results, state some key conclusions. So these things [00:32:00] are, I think already, possible with what we've seen already with LLMs and gen AI. I think it's. a short matter of time before simulation practitioners actually start making these kinds of features available and we start to see value
John: Nice. So Dave, we have time for two more questions. one with the different areas of simulation that you've been working in. What new areas do you think simulation can be applied or what new types of simulation, do you think might gain more traction into the future that make you really excited? second to last question.
Dave: I do get pretty excited about the use of AI in simulation, because I think there is the potential to very efficiently, add a huge amount of value by, leveraging AI to automate. basically to make good predictions about some of the situations where simulations, struggle and, or, the exact right parameter to use isn't known, but, now [00:33:00] you train an AI and you have, very fast, efficient way,to get a good value of that parameter based on what's going on.
And you just, there's so many ways in which this can be folded in.and the ability also to train use simulation, to train an AI on a specific application, a specific type of problem, and then get very fast, very inexpensive inference, as a prediction, right? it's like instantaneous prediction.
And I think what's going to happen. so I know some people have said, isn't that going to cannibalize simulation? People will start using. training models and using AI inference to replace, simulation to get, the same prediction. But I think that's not what's going to happen. I think that it will cause the use of simulation to drastically increase because the value will be so high that people will want to train [00:34:00] a, AI models across a huge number of different, use cases, applications, specific types of problems.
And lots and lots of simulations are going to be run in order to have the data necessary to do the trainingand then the value will be like really efficiently and quickly realized from those trained models. And I do believe AI is at the point where those train models will be very accurate and as good or nearly indistinguishable from having run the simulation.
from scratch.we're, I think the industry is going to get very good at that. And we will be able to offer,Instantaneous prediction from a trained model. So sure. You pay something up front,now imagine all of the end users. and then I was able to use that model for effectively instantaneous or near instantaneous prediction.
So I think this is going to impact, all areas of [00:35:00] engineering simulation and all different applications. So I think that's coming and coming soon. and as far as, new areas, I think, places where we've seen simulation. very often get left out is,life sciences.
So biological sciences, cause they don't tend to be full of the same sort of computer science type folks that You know, tend to be able to pick up the simulation expertise.and so this is where accessibility really matters. we need to tailor user experiences that work for, life scientists,healthcare clinicians and others that will benefit tremendously from simulation, but it just needs to be exposed to them in a way that makes sense to them and is usable.
And then all of a sudden you're going to see, I think hundreds or even thousands of new simulation applications. With, some pretty rapid adoption as long as, these professionals [00:36:00] are able to use it effectively and we don't bog them down with a bunch of, stuff that they're gonna have a hard time dealing with,the simulation didn't converge or, you need a finer mesh or all that stuff, right?
We've got to automate all that stuff away and give them something that just works reliably. And I think that will be the key to unlocking the growth of simulation across the board. A broad set of
John: Nice.
Dave, last question, for those people who are interested in getting into simulation. So there might be some students, high school students, undergrad students, listening to this podcast. What backgrounds do a lot of people that do well in simulation, what backgrounds do they come from? is it only computer science?
Is it math? Is it social science? what common themes do you see between people who, between people that do well in the field?
Dave: Oh, great question. sure. physics, math, various engineering disciplines, are a great background as well as computer science. there's a lot of other, [00:37:00] skill sets and talents that are really important. I've mentioned design a few times.I think design is critical to the success of simulation.
So people who enjoy, user experience design for, complex problems,might really like a career with a simulation company with a software company. That's focused on, simulation and,all the things that go along with design,really good user interface development.
so I think there's actually a wide range of. skill sets, that,we'll play a role in, and we'll have a lot of value. And of course, it doesn't hurt to enjoy, engineering and,those kinds of problems and applications could be any type of field, aerospace, chemical, biological,very often that's, what the simulation is being applied to, is to solve problems and make predictions in those areas.
So those are all, all of those different sciences and engineering fields, I think, can be a lead in [00:38:00] to simulation as well as, some of these,I would say design and experience related, interests, it's, simulation casts a wide net these days, I think,
John: Cool. final thing, Are there people from your past that helped you get into simulation that you want to give a shout out to before we wrap up?
Dave: oh, wow,my old PhD advisor, Kim Mulvig,was a big influence on me, our chief scientist at EXA, Hu Dong Chen,just a brilliant, Humble, wonderful guy that, had a big impact on me.and,more recently, I w I work for a guy named Don Ferguson at Ansys, who, I would say has been a great mentor and really had a big impact.
our excess CEO, Steve Ramondi, we really went through a lot of, simulation thought process, how to build the business and how to make simulation more accessible and how to drive adoption. So I really, I'm grateful. To, to Steve as a mentor. yeah, I'll, I'll stop there, I've been very [00:39:00] fortunate to have worked with a lot of, really tremendous, brilliant people that, I feel like I've really learned from
John: Awesome. Dave, we're grateful to, have the opportunity to work with you, with the epistemics team. And, with that, I think we'll wrap up. So thanks for tuning in today, everybody.
Dave: my pleasure, John. Bye everybody.