Hello, I'm Rob Hirschfeld CEO, co founder of racket and your host for the club 23 podcast. In this episode, we dive into data operations. This idea that with all of the work we're doing in AI and ML data analytics analysis, that you actually have to steward your data, you have to be able to have processes controls, a lot like what we have with DevOps in infrastructure, but similar types of concepts, governance controls automation, around how your data is flowing in your system. And so that was the topic that we took on in this discussion, I know you will find it fascinating.
You think the Cray acquisition plays into this stuff? Just that wasn't I mean, I'm thinking about that as a proof point for what you're describing. I'm it, maybe we could have a whole session about quantum I mean, maybe I'm just not tracking the right channels, I haven't seen quantum as that material of
the I'll send you a few links from a few vendors, websites that you'll be very surprised at. Okay. And not not not not not that I'm boosting the country. But we have two innovation clusters in Quebec and Ontario, both gear to AI, and one in particular to quantum, we kind of lead in that area. And the two are cross pollinating each other all the universities in Quebec in Ontario, BC, but there's a lot of startup action. And there's also a lot of known companies that are taking advantage of that. And they're all funded, they're all getting funding from the federal government. So they're very big strides.
Or they're pretending like it's
it's a country wide initiative to lead in both those areas of technology.
I just and the app, and the applications that are surfacing for it. I guess that's a I mean, what what are they? What what are what are people surfacing as utility for quantum is just performance encryption, statistical analysis, maybe I just don't,
it's it's encryption, it's statistical analysis. It's Wow, how do I describe this? In a nutshell, if you're sitting in a lab environment, of any kind, whether it's, you know, for a particular hard, good like car, and you want to reduce the amount of lithium that goes into the batteries, let's say, for cost or weight, use quanta, wherever you can. If you're University oriented research for healthcare or drug discovery or anything like that, you would use it the same way. Because the amount of computations that you can do per millisecond is X times what you can do today on other kinds of computing. So it's early days. There is a lot going on, believe it or not.
It's some are bad, like, risk versus Sisk. That with quantum, you can do some domain specific computations very, very fast. Right? It's just not a generalized application yet.
Right. Are you are you seeing a lot of use of quantum against fully homomorphic encryption? Mean?
I mean, that that's, that's what drove her like quantum. So yes. Well, the
question is, is it continuing? Is it making it out to the, you know, into production and being used? Yeah. In any particular fields? Yeah.
There's a pilot ongoing right now that I can't name the players. But I can tell you that it's a major health care provider and it's to encrypt test results and all of the PII for healthcare.
Project. Yeah, I'm sorry. It's a massive project. Yeah, yeah. Is it? Is it primarily? Well, two areas that seem to be kind of that were FHG makes it potentially makes a big difference is in the whole collaborative area where you've got multiple players consolidating or compiling their data from a variety of sources, but only in under the condition that it remains encrypted. And there there is a is a real prohibition from anything that would leak that information. So it's not just maintaining PII, it's, for example, in in, in insurance, all the insurance carriers are obligated to benchmark themselves against everybody else in their, in their industry, their industry segment, that by regulation, the problem is, ie they're also prohibited from sharing information or inadvertently leaking information to one another, because that starts to look like cartel behavior. So they're in they're in, they're in a, they're in a catch 22. So fhe has been one of those things that they continue to look at and look for, but I'm not aware of any of the any of the insurance industry folks actually running a significant project. That's why I was asking Joanne.
Yeah, no, well, okay, so let me make this localized. There's three projects ongoing, individually, and grouped. One in Europe, one in the US, and one in Canada. But the Canadian one is sort of managing the whole kind of thing, because it's a public private government partnership. So public, as in hospitals, and health care, government oriented and government funding, private because there's private corporations that are involved. And academic as well. So IBM has a piece, there's other players that are in there as well. One of the big farmers is involved also. So they basically the way the structure is oriented is there's a Canadian arm, there's a European arm, and there's a smaller pilot going on in the US
to avoid the normative norm of a an over of an umbrella organization.
Now, or just a collaborative, okay. It's, it's a we need to do this, we because, you know, at some point when it comes to plenty of war when it comes to AI, Canada, and the US will have to agree in some way. And we're all masters under the EU anyway. We already have initiatives that have been publicly declared, declaimed for innovation centers, and for policy around AI. So we're kind of ahead of the curve there. And that's part of why Canada is participating. There's so much going on, it's, it's really hard to put it all into short words that I can actually disclose. Because as you guys know, I work as an assessor. Part of what I do is in assessment projects, okay. So but the commercialization is definitely there, the encryption part, which I there's something that I can cleanse a little bit and then sent to you.
Yeah, I'm pretty aware of what's going on in Europe with FHA. But specifically, and I'm fairly up on what is being done commercially, with conventional computing. Basically, with hardware assist that goes into you know, a backplane somewhere, but yeah, not not clear to me what's happening in quantum I'm just not in that loop or not. So
well, you know, I mentioned to you guys that I was just, you know, in the process of doing my course with IBM, on quantum, which is really interesting, by the way, really, really interesting and to Your question rather than go to market and moving forward think chips that are not made by Nvidia?
Right. Sure.
It's a
I mean, there's so much new hardware stuff coming out. I think people underestimate the how disruptive the hardware market is going to be from that perspective, especially
with RISC five.
I see. Do you see it coming in at the server level?
Yeah, I do.
Kristen, we're interestingly enough, I feel like sometimes we're we see we're, since we're so enterprise, we were sort of slow adopters. I don't know how to get maybe this is a we just have to fund this. Get involved in in more of the esoteric, new new model stuff, because we can be very helpful with that. Yeah, okay. I think that we could it's we right now we have a tendency to work where there's no more established product, even though we don't have to be they're just a matter of where they were where the bills get paid. Nobody nobody pays for to automate prototypes. Even if they should be I
feel like we've left poor Dave out in the cold here. Well,
I and our topic our topic for today is data cloud. And and all the all the data ops. And this was this was a go back to data ops for the year, which is I know, one of Dave's favorite topics here. And so so we got we actually got pretty far down in this. We were talking so it just as a reminder. I've been I've been talking through this sort of data DevOps concept. And what we what we got to was this challenge with there's multiple layers of data operations that I think is confusing from a market perspective. My dogs barking in the background, I don't know, hopefully, that mic screens it out. You can't hear it. Yeah, we don't hear it. It's amazing. But but there's there's an element of building clusters and managing clusters and doing the operational work, which is ones one side that I think when we think of DevOps, that is a place where people think about it. But on the other side of that, what y'all keep bringing up that I think is surprising and interesting. Is all of the work necessary to manage the data govern the data and provide that type of capability. Which has its own, you know, automation requirements, its own governance, controls, security, pipelines, all all of those things. Are those two just totally different topics? Or is there some, some convergence?
Well, a lot of what's happened in modern data stack and the data engineering, which overlaps? I mean, it's hard to distinguish when somebody says, Oh, I'm, I'm, I'm doing data engineering, I'm okay, you know, fully, I'm pulling together, all of the piece parts putting together DAGs. I'm running the pipelines. And to the degree possible, most of the most of the technology that I see being used in modern data stack, is
trying to abstract, a lot of the underpinning that you were just mentioning. And just, you know, I don't want to see it. I don't want to know about it. Don't Don't bother me. It's it's almost as though there's a data engineering crew that is living above what you were just speaking about, and delivering to the data scientist and the analyst and so forth. All of the all of the goodies, all of the data, lakes, so forth, the platforms they don't touch, as well. They're not doing the the analysis. Maybe the only point in time where they actually get into analytics is when they do reverse ETL I watched the data pipelines, pull out their operational data and feed it back into their own processes in order to optimize them. But they're not. They're not doing feature engineering, they're not been doing major machine learning for a lot of these things. They're simply enabling it. So it's almost it's a, it's a three layer cake.
The weird the place that seems strange to me is that my understanding was that to do the top level stuff, they actually had to do a degree of like, firmware and patching on the GPUs, or the specialized processing chips, like the firmware like the the algorithms, and the firmware actually mattered at at that top level. Which is why I keep expecting somebody to be like, Oh, I care about more of the operational components. But if that's not happening, then, you know, the, the platforms are
so many, I mean, are you talking to data scientists who are getting down into the firmware layers?
I know that they are the I guess not. i It's funny, they don't I don't know, when I've talked to them. They don't have very crisp definitions of how they're interacting with these pieces. Usually, to your point, they're they've got a platform that they're they're queuing jobs up in a platform and they run the jobs. They, you know, they they're faster if the clusters bigger, but they're not because they're different.
It's because they have different skill sets. somebody that understands running infrastructure, and operating infrastructure and doing updates and things is not the same skill set as somebody doing deep data science. They are. They're very different skill sets.
But don't they need specialized? I mean, the thing I heard is that they needed was they being the data scientists, so I was at the NOC data science, this was I was at a high frequency trading show. And they're consuming all the GPUs and high frequency systems, but they're very concerned about firmware updates, firmware embedding, specialized, right.
Quantum the high speed concert almost, almost in a class by themselves. Okay. Yeah, I mean, there's so so tuned, you know, they're, they're so focused on you know, squeak performance, and it just, they they do, you know, crack crack the box open. I don't know too many others, even even in the, the pharma areas where I'm pretty familiar, the data science that's going on in pharma doesn't generally have to have to go that route.
Um, I would beg to differ slightly in the labs it does. And the place where you find that mixed skill set is in manufacturing, in automotive and aerospace, in defense in high tech, electronics, manufacturing, anywhere where you have a need to be able to mix the two together, because your performance will change on FPGA, your performance will change on where and how software is embedded, ie prom level. Who takes
joy? Who takes responsibility for performance? Is it the data scientists? Or is it the is it the data engineering? Is it? Is it infrastructure? I mean, the data scientists are the are the top top tier guys are going to be sitting there banging on everybody that says, you know, don't spend any more money, get these things through faster. I need performance, you know, squeeze everything you can. But is it the data science folks that are that are that are cracking it open the way the people in finance or or because I'm not, I'm not seeing that.
In some cases, it's it's not. You know, you can make one comment for all it varies by segment. It's like in electronics manufacturing when I was at Celestica, or Intel or any of those kinds of companies. Yeah, you got data scientists cracking those chips open and sitting with FPGA guys saying make it do X or, you know, increase the throughput? Or let's go to another nanometer chip? Or how do we how do we, you know, make the silicon, more performance on the heat side, and heatsink, you get a lot of that electronics manufacturing, which is usually driven by any of the 10 industries that if supplies, so think spatiales. Think robotics, that's where those two skill sets come together in heavy engineering environments, where it's driven by customers, more than, you know, here's the product that I offer. That's where those things did come out, percolating up.
I can, I can see that. But I think in the average enterprise, it's, it's the kind of opposite, you're not going to see, you're not going to see that it makes sense. If you're an electronics manufacturer, or a chip manufacturer, or you're deeply involved with hardware that you one would, would want to and focus on hiring people with that kind of mixed skill set, and that you would care about those things that that adds up to me. But if I think about, you know, I don't know I'll target. If I think about target as a company, their data science team is not going to be cracking anything open. They're going to have a separate infrastructure architecture team that's going to be focused on on that versus
an engineering group. That's right.
And we had that even at GE, you know, on we had, we had a team of data scientists that were very much focused on tuning machine learning models, focusing on how to optimize to solve specific problems in a in a domain space. And then you would have, you would have the data engineers that were focused on how do I get these datasets into a suitable form so that they can be consumed by the models that the data scientists are using? And then you had the general architecture and engineering team that was worried about? Where do we run these models? How do we store this data? Where do we keep it? Where do we move it all of those things? And that was kind of the general separation of concerns? And that's at least held true at the other places that I've been involved in? In what I'll call the general enterprise?
Yeah, no, you're right. I mean, outside of the big ecosystem called manufacturing, and it is a big ecosystem. I think the only place that it would probably be absolutely germane would be in pharma, health care, or life science, pure life science. So pure research, drug discovery, and lab clinical and all of that, and maybe in like in hospitals,
do you think that maybe changing with some of the work with some of the cloud providers I know, at reinvent this year, there was a lot of talk around pharma and healthcare and specialization around drug discovery and such? Yeah,
because drug discovery, you know, drug discovery is not only high performance, compute and ripe for quantum, but there are med devices now. And you have, you know, think more from the consumer side that will absolutely need that kind of mixed skill, looking at it and working with it, whether it's, you know, anything like and I'm sure you've seen the commercials for them. In the US, there's now an implantable device to for sleep apnea. Right, that frees you from all the gadgetry that you had to wear? Well, I
haven't seen that. Sounds interesting.
Okay, it's been it's being advertised or E skins, which are thin layers of skin, not human skin, but a skin that's completely a computer, you can scroll it, you can do whatever you want with it, any of those kinds of applications and even some of the glasses that are about to be announced. Excuse me, as well as encourage telemetry you need those kinds of capabilities together. So that's where I think data ops, is we're gonna be up all the way parametrics we're.
So being able to mix all these types of like Nevermore We're talking about combining data as a specialty, not the actual mechanics of the systems behind the scenes. Okay.
I still question how much we will see a lot of these functions, leveraging AI as a way to automate even further.
Well, if you question your question, its adoption, or you question its ability to address it. I
question the timeframe, how long until we see this occurring? Because
I don't think this issue, is it a skills and background issue?
Right. Well, and that's, and I think that, I think that it is difficult to find people that are good at being able to do some of these things. And, you know, if you, if you're able to analyze a, you know, the heuristics of a data set, you can get probably 90% there without a human. And maybe you still need that 10% For a while, until, until things can be trained to an acceptable level, but I see a lot of those things being automatable.
Even, that's where the data ops stuff comes in. Right? But where do we even have the tools for automating this? Like, I guess this is where we're back to, there's, there's process and people go ahead, well, what tools,
what tools, etc, is lacking? I mean, with with Python, you can do a lot of data manipulation now with Python, and the libraries that are that are available in that ecosystem. And if you've if you've ever asked Chet GPT to write Python code, it doesn't do a bad job. Is it flawless? Is it perfect? No, it's not. But it does a pretty good job. So maybe your maybe your data, you know, maybe your data ops team would have been a team of, you know, seven, or nine. And now it's a team of three to five, something like that.
And the eight, the A using Asian tree, it with with a lot of the, the new tools, the quality of the code, I'm getting out of these things for Python, where it might have been, you know, first run, maybe was an 80% solution, and then you had to massage it and run through again, it's coming closer to 94 95%. And, okay, that's pretty, you're you're actually finding a lot more people who are using Python, as kind of a use Python and API's as a means of doing data pipelines and building them without this without the same same kind of tooling folks has had to use in the past. We're leveraging
to see your See, you're suggesting that the data ops, here is Python, I guess I'm looking at is there the equivalent for an Ansible or puppet or chef thinking about data ops? From a right from DevOps for a data ops perspective, where there's a generally, the tool set is broken down into understood components that have a higher abstraction, right writing Python code is feels early to me, and maybe I'm misunderstanding it.
The thing with Python is you can weave together virtually any of the existing tools, whether they be open source libraries, and they don't have to be Python written, they can be Java, or rust, or pick whatever you like. And you can also use all of the services API's that exist out there. And because all of these things have been documented, and most of it has either been slurped up by by the API providers, or, or you can simply point it to where the documentation is. It can build you a functional, I'll call it a prototype of what you're looking for. I mean, maybe spend an hour maybe that's it and then you have something that's already got you as Rich was saying 90 plus percent the way there then it's more of refinement, tuning optimization and such for the year specific use case. I mean, I don't know how much more you really need.
Yeah. And then with function calling being available? That's, that's a big deal. I mean, the question is, Do you need something that's an analog, as you were, you are thinking about it, Rob? Yeah. Or do you take those existing piece parts? And, you know, I hate to use the word own, you know, overlaid, but it is, in a sense, a kind of, you know, let let the let these kinds of intelligence is right for us use the deterministic aspects of those of those tools and, and do the job for you.
For what Joanne was describing, where you have these different sources of data, you're, you're manipulating them, you're coming out with a result, or you're creating a feedback loop? Is that does that just become a custom application? Or is there a standardized pipeline for those pieces? Is it just Python Python code, that you're slapping together?
Our Python code, it's there are tools and then there, and then what you're doing is you're making use of things. I mean, at a slightly different level, but what Microsoft has kicked out called autogen. Yeah, basically creating specialized agents that literally go into a chat room. And you've given the the very the high order tasks and set them loose. You give them a budget. And it's, it's pretty it's it's not as it's not as deterministic, it's not as as rigid, methodical, methodically rigid means of putting all of this together, but the outcome when you watch, and kind of review this transcripts of these agents talking to one another, and figuring out what went wrong and how to fix it. It's, it's, it's spooky and exciting. Yeah, yeah.
It's like watching console. Ai. If you've ever used that, it's another open source. And you can download it and run it. It uses open AI API's, and it does the same thing. You give it a task, and it has a team of agents that will outline the task that needs to be completed, it will assign specific tasks to specific agents, for example, you sat want to write a Python Hello World app, it will outline what's needed, it will assign to the programmer, the task of writing the Python code, you can watch it write the Python code, then it will execute the Python code to see if it actually produces hello world or not. And if it does not, it will go back and suggest fixes to the programmer. And the programmer will region it will do all of this by itself. You have
not an anomaly. The Pro you're talking about the programmer agent not. Auto Gen. Auto Gen is is also you know, it's Microsoft. And they have they have integrations into open AI. But you can also build it on the back of open source A, A is sure
like llama or whatever monitor
what you're describing to me is still not using somebody's data here, what we're you're describing is LM directed agent activity, right? Are we oh, what's the next step? Okay.
It will be if the problem space requires you to put together you know, a big data frame, for example, and then test the data frame, either for accuracy. Look at its performance. Those are absolutely within the realm of of durability. And again, it's pointed at it you know, at the the aspects that you're interested in optimizing, protecting, demonstrating so far If it's it, it actually kind of moves away from what human beings like very often, which is a cookbook, repeatable kind of an approach. And as long as it can be as long as it's observable and and comprehensible to somebody reading the transcript, then you know, the question is alright. Are there are there ways of thinking about it that are that are different.
mean, you can assign to, you could have a, an agent that is an expert on your dataset. And you could say, Yeah, this agent has full access to my data set, it is the expert on the data set itself. Any of the other agents can can query it about specific details about that data set.
So you're adding it adding, and it basically, it's a geek LM gatekeeper, where it's an agent gatekeeper for data that is closer to what we're talking about with data ops, where yeah, we're, we're saying, Alright, I've got, you know, and you've actually managed, I've got to manage my data, somehow, I've got to secure it, I've got to protect it. What you're actually suggesting is there's an API gateway that allows you to move from a higher order, you know, query whatever raw data and actually say, here's an IT here is a system that is, you know, going to answer your questions for you. I
would argue it's not a single agent, I would argue it's a collection of agents or a team. And your data ops team is really a team of agents that have skill specific. That's right, with specific skill sets to address the overall needs against that data set. And that theme of an agent can address whatever your needs are.
Yeah, it's a swarm. Yeah, yes, we're, yeah, it's a square. So there
is an enterprise somewhere, building, basically taking their data set in, you know, in a data lake or in some on some platform, and then you building a platform on top of that to be this intermediate? I guess, no, this is that because we have to operate. I guess there's a premise that I have that you're collecting data, you have to operate the data operationalize the data, right. And then you have to provide access into that pool of information somewhere to make it useful. That's to me data operations. Well,
you do and at least the trend for the last, I don't know, since we all heard the term big data, which you can imagine how long ago that was now. There have been efforts you can look at, we can go back, I'll throw out a lot of terms data mining, data warehouses, data, Mart's data lakes, you can, you know, we can go on and on. That's right data, yes, data, lake house. All of these things really boil down to the enterprise trying to wrap its arms around what to do with data. And the problem that every one of these things runs into is how do you properly separate data that you want available to other groups and departments versus data you do not want? So if your HR, you and payroll may share information about compensation and such? Do you want that shareable to other departments? No, no, you don't, you don't want that available in HR might have access to benefits information that should not be accessible to anyone else, including payroll and, and so all of these divisions of function, and such necessitates some boundaries in that data sets. You can't just as an enterprise, at least an enterprise of any real size, just throw all of your data into one gigantic database and say, everybody can just do whatever they need to with all the data. It. It seems nice, but that's right. It doesn't work. And it's not practical. And so everyone struggles with is how do you how do you create the right security and and, you know, domain boundaries within that data so that you can leverage your data. When it makes sense across metrics. And just
like any human team, what you do is you establish policies, then you stablish and you, you basically enforce those policies at the at the interfaces at the API. We are really down at the API call. You know,
this agent Sorry. Yeah. No, no, I, I guess, is there a tool or platform or something that people are turning to to do that work? Or how are they building that? Like?
I, I think maybe I'm being concrete but Well, I mean, I'll tell you, they're their platforms where people are trying to trying to achieve this. You know, that's one of the big things that snowflake has done, as an example, is they've been working towards trying to achieve that kind of thing in the enterprise, with an entire ecosystem. So it's not just the platform, it's the ecosystem around the platform. And, you know, is there a final solution? Not yet? No. But there's work, there's work being put into trying to kind of crack that nut, you're,
I mean, just from walking around, like reinvent and things like that what you're describing is, there's several platforms that are trying to be the definitive data? Or they're, they're trying to be a data platform from that perspective. And I guess,
yeah. And remember, I've been I've been, I've been busting my pick on that very problem for the last three, four years. Yeah, this is this is why I think Exactly, that's exactly where we are. It's fascinating
to me that we've we've, you know, I know, we want to call them data lakes, it sounds sounds like their data puddles. But I mean, there's we're stuffing data inside of mattresses, and anything out how we're actually going to pull the data out and manage it and make it accessible is it that's the operational challenge with this, right? It is, and
a lot of a lot of the things that also go or need to be taken into account that are operational, and these are data ops, this is my data husbandry soapbox. One of the things you want to do is make sure that you don't have people just throwing the same data or variations on the same data into the same data lake and then trying to figure out, which is the authentic level, you know, the off the authorized version of a spreadsheet, for example, or a document. And so data deduplication, all these things, what it, what it means is, you're generating a shit ton of metadata on every piece of source data that you put into these collections, an
AI is going to only amplify that.
Well, AI is probably the only thing that can actually make use of it. Right? So we're back,
we're back to the metadata, the metadata being part of the
alright. It's time
I feel like I've been asking the wrong like, I'm asked I, this is this is I feel like I keep asking the wrong questions. And I'm not I'm not connecting with with, there's something I'm missing and I'm in the middle of all this stuff. So but I appreciate you're clear as mud? Well, I have a question I might need, I might want to draw like a picture or something because we keep trying to do this with words. And I keep I keep feeling like that they were their stuff moving. There's there's pieces and parts moving and there's there's work to husband, you know, the husbandry pieces of this operations components. But I don't I don't know where that's it's occurring in bits and parts. I don't feel like it's occurring as a discipline.
Not Yeah. I don't I don't think you can know, we're, we're talking pretty much at a, you know, kind of the edge of the, the edge of the known world
here. Seems weird for something but people have been collecting data and working on and doing for for as long as they have been.
They've been collecting data. But you know, Dave talked about big data. Yeah, fine. You know, what, 1520, almost 20 years ago, did you start hearing about that? But it really wasn't until maybe six years ago that you really had something that even looked remotely like an engineering design. Applying for data, we've had software engineering for 20 years 30 years,
well, then that would follow that would track for us missing the operational piece because it took a long time from the software development or an operational
My point exactly. And, you know, the speed with which all of this has happened is, is pretty fantastic. But it really has only been seven years than we've had anything that even remotely look like data development, sign language, data engineering, that was truly distinct. And, and a focus that data DevOps organization could address and put their, their, their arms around.
I find that really interesting, because when I was at Gartner, and even before that, before the words, Master Data Management actually became public knowledge. We were doing it then. And I don't know what the but but software was software in those days, and then Master Data Management erode, literally came out of manufacturing, because we couldn't figure out how to make the round peg in the square hole, you know, actually equal in some way. And we went,
I remember talking to folks about, about, you know, the software guys that were doing a lot of this work, and you kind of say, Okay, you're building all of these, you know, you're building networks, you're building, you know, all of this technology, all of the software. What do you think about or what when you think about it, what's going on inside the, the freight cars that you're moving, and they basically sit there and go, I'm just moving megabits, man. And that's not
right. And so we looked at it from the point of view of, you know, the blob, and if we go to Orme, and we use the blob, we can make blobs all coagulate into a data pipeline of some sort. And,
by the way, data management was a was a very useful, very useful concept and very useful, kind of set of structure actives on, on on data. But it's it was, it was, by comparison, rudimentary to what had already been developed in the software in the software community.
This is one of those topics that we are going to keep coming back to. It's stunning to me, the amount of effort and time we spend talking about data with out is clear a boundary on what we're doing to govern, manage control over this data is clearly a major product category coming up, it's a place that if you're not struggling with it, now, you will be soon. And I'm interested to see if something akin to the DevOps manifesto actually surfaces out of data operations, although perhaps we're not actually talking to the right people, or with the right groups to see if there's a convergence coming together. around that, that would be a fascinating thing. And Cloud 2030 could be the place where we start doing that type of work. One thing that didn't make the cut on this video, so I encourage you to come to the meetings, we actually had a really fascinating conversation about quantum in the pre part of pre recording part of this meeting. And we will come back to quantum and talk more about if you are interested in quantum or have views in questions. Always, please feel free to join us at the 2030 dot cloud. We'd love to hear what you have to think. Share your questions to.
I'll see you there.
Thank you for listening to the cloud 2030 podcast. It is sponsored by rockin where we are really working to build a community of people who are using and thinking about infrastructure differently. Because that's what rec end does. We write software that helps put operators back in control of distributed infrastructure, really thinking about how things should be run, and building software that makes that possible. If this is interesting to you, please try out the software. We would love to get your opinion and hear how you think this could transform infrastructure more broadly or just keep enjoying the podcast and coming to the discussions and laying out your thoughts and how you see the future unfolding. All part of building a better infrastructure operations community thank you