From Melancholy Metal to Data Products: Sisu, AI, and Purposeful Data

Speaker A: Welcome back to the Jam. In this episode we're following a journey from the land of a thousand lakes, Northern Lights, the home of Santa Claus to some at least, but also Tuevo Kohone, the famous Finnish AI Professor Nokia, Angry Birds, the Leningrad Cowboys and VI Kate the sad melancholic typical Finnish music to the land of endless dunes 1000 one night tales shakes desert cities and high tech Oasis Abu uh Dhabi It's a massive thermal shock moving from but as our guest today knows, Sisu, that unique finish grid travels well. However, Sisu has a side effect being hard jaw. It's that special kind of stubbornness where you'll refuse a shuffle while chest deep in a snow drift because you've decided to dig your way out with a spoon just to prove a point. That's exactly the energy needed to standardize the world's data. Today we are joined by the igniter and lead of the Open Data Product specification in the Linux Foundation. The Dr. Jarko Moilinen, Department of Governance Enablement 1 of the Emirates, the product head of a data platform, AI and data and this is his life besides the Linux Foundation.

Speaker B: Jaako, welcome to the show today. Please tell us about yourself in the beginning and how that hard jawed stubbornness Robert was talking about in the beginning is serving you in the Abu Dhabi heat.

Speaker C: Yes, thank you. Welcome Jarko Mojlanen, which is very hard to pronounce if you're not a native Finn. So I accept all the forms and I have heard all of them. I've been living in Abu Dhabi for the past four years and I'm proud to be part of the building what aims to be the world's first AI native government. As said, originally from Finland, but today my life is in UAE Abu Dhabi. That environment creates a different kind of ambition. It puts pushes you to connect the data and AI directly to real outcomes. So my focus is on turning data uh into miserable business outcomes. I work with data products, not datasets. I don't work with dashboards. I work with product with a clear purpose, defined users and expected results. This role that I have has given me the opportunity to combine the two worlds, data products and AI uh into solutions that actually create real value. And as said, I also lead the work around the Open Data product specification under the Linux foundation and build also platforms that make it operational. So it's a kind of combination of two different worlds. This hardshot mindset actually matters. I said this standardization in the ODPs it takes time. People resist the structure. You have to stay Consistent, you have to push until it works. That's what uh, Diesel is in this context.

Speaker B: Thank you for the introduction and also telling us how to properly pronounce your name. Can you tell us a little bit about your way. So how did your journey plan out from your PhD to now being in Abu Dhabi and as you said focusing on the data products?

Speaker C: Yes. Originally I was for a long time leading data platform development in Finland in one specific company, in specific sector and uh, even that was the point when the odps actually was for the idea. But then uh, I kind of found myself bored in Finland. I didn't get enough challenges and I'm kind of a uh, big problem solver system, thinking person and everything else. And then I started to look out at okay, where in the world do we have something interesting going on? Then I kind of stumbled on the Abu Dhabi and uae. Big goals, big ideas. And then I thought about well why not? And then I agreed at the time CEO that let's make a uh, path out of the company for six months and then I will leave. And then I ended up in Abu Dhabi in a completely different world, building my life again on a different culture. But then I came here for two years but I'm still here after four years so I kind of sticked here. So this is where I found my new home.

Speaker A: This is very interesting. I think so. But you went to Abu Dhabi to work with this data platform? I guess. How did you move to this Open Data uh, as a product idea where that thinking come from and when did you decide to move that to the Linux Foundation? Because that is recently isn't it? That's after you moved.

Speaker C: That's correct. The idea of the Open Data specification was born in Finland in this while this one other platform was in develop and uh, we found that the DCAT dataset definitions and everything else that was uh, on the market available was not actually fitting for the purpose. They were not business oriented. Our platform was thinking data products and they always have a business intention. And I just had an idea and I suggested a goofy idea in the company. Let's not take it too seriously, let's define one. And we started doing it inside the company. But then soon I was discussing with my CEO that hey, this whatever we have been defining now could actually be useful in bigger context. Can I spin it out of the company? And he said go, go ahead, do whatever you want with it. And um, as long as we get the results and the benefits. That's how I did it actually spinned out from Controlled, limited environment to be an open project, open standard. And it started on its own life in on their own domain which I purchased for it and started developing with a couple of my colleagues and business uh, partners that continued for a couple of years and more adaption came along and I saw okay, there is traction, there is more contributions. And then I was thinking m, we actually need a good home. We need something that actually acts as an umbrella for us. And it also most likely brings some credibility and other benefits as well. And that's when I started to approach the Linux Foundation. Coincidence? I happen to know some people already from the Linux foundation so I wrote an email to them directly that okay, I have this thing going on, can we discuss that? I actually moved this all under Linux foundation and evaluated everything and then eventually everything moved under Linux foundation somewhere around 24. Yeah, it's a long journey and there was a specific reason why we wanted to go under Linux foundation and it has been serving the purpose because it really increases the credibility of everything. And as you might know, Linux foundation is now full of data and AI, uh, related work, standardization and everything. It has become in my mind the hub of this kind of work and as an umbrella more than some other standardization organizations and so on. So that's the story.

Speaker B: Okay, thank you so much for giving us that introduction. So in recent episodes we already talked about Open Data and also fair data principles. Maybe can you tell our listeners where that openness in the data products come from that might not know the specification as much? Can you give us a short introduction and what does that openness mean for you?

Speaker C: Yeah, something that some people confuse when they hear the combination of Open Data product specification. They might be automatically going towards open data. This kind of a very limited concept of different uh, data and so on. That's not the intention. Why there is the word open is that it's open as a standard, much like the, in the API world people might know the uh, Open API spec. It's not about open APIs that anyone can consume, but it's about the openness of the standard. Uh, so we kind of adapted that. Okay, let's put the open over there because we want to communicate that this is an open standard. This is not something proprietary and given to you as is. But this is open development, open standard and now that we are under the Linux foundation it should be very obvious. So it can be used in open data but also on commercial data. So it's not limiting the use in that sense of your products at all. So that's the Story short on that open aspect. And of course the value is that I get a lot of contributions, I get a lot of ideas because people in the development world also they love everything open because then they know that it will be shared forward. It's not limited to me or some

Speaker A: company own and it is standards then that you can trust will be there in a few years from now because you can always go back to the specifications, they're open and so what you do is available also in the future. I like that one. Are you also supporting with code for these open standards?

Speaker C: Of course. I have to think about my time and the time uh, what the people around me as well who are contributing to the specification as well. But I do code and I do also develop on top of the ODPs myself. Complete platform has been recently built on top of it to actually make it real and actually show people what it can do. Because I have found out that people don't understand unless they see it. So because all the options, all the benefits, values that you can get from applying the ODPs is not easy to see as a specification. So I did with the help of a team, build a complete platform which is bringing all these good sites up. Um, but of course other people are developing on top of it different Python libraries and things that how do you actually programmably consume and modify and every validate the ODPS spec as a result and so on. The code is part of my life, I've never got rid of it so I want to do it as well.

Speaker B: Okay, is there also when you talk about the openness and the specification, the standards then other people develop on it? You said a feedback loop or how are you integrating different communities feedback into the standards to make it even better? Um, progress in the future.

Speaker C: Yes, standard and all the versions and all the decisions that we do are Openly recorded in GitHub. So there is a GitHub process where anyone can go and raise an issue or a new idea or have a conversation with the community around the ODPs. And then we also have WhatsApp groups and other kind of uh, own specific groups for specific discussions. For example business people. I found much easier to get in a WhatsApp group and uh, share their uh, business things around the ODPs and then I can hear it from, directly from them TTO levels, CEO level people, they don't go to GitHub very often. I don't think so. It's very developer oriented, technical oriented, but that's where the results quite often come from.

Speaker B: There.

Speaker C: But that's not the only way. They come to me with lingering messages. They send me email that the channels are endless where I'm getting all this feedback. But it also requires that I cannot just sit and wait. I have to be active as a maintainer. I need to communicate and I write blogs and we have a block in medium for the ODPs itself. And then people get inspired. Oh, uh, okay. And it's not only about what is going to be visible, but I also share what I'm thinking, what I'm planning, what's on my head. So kind of uh, having the multi channel discussion and the ideation happening around me. And my job is to get people engaged into this development but my responsibility is to take the ideas and see the ODPs. How do we actually make it part of the specification? So it's very hard to misuse but still easy to use. So that's my job.

Speaker A: I guess. You get a lot of feedback from the whole world. Uh, so people from everywhere, North America, South America, Australia, whatever. How do you tackle that? And I think it must be a lot of fun as well. You will get a lot of cultural differences as well and weird questions. How is your life in that sense? Can you give us some examples of what happens then?

Speaker C: How do you find this kind of uh, as it is not something that I get paid but I do it out of passion and because I have the goal. The odps has been in my head. So I do it of course very eagerly. But you're correct, I get a lot of feedback from Canada, from the U.S. and sometimes it's coming from Brazil, sometimes from Netherlands, Germany and uh, quite recently from India as well. And then couple of people very actively on Australia. So yes, very global is my community and I live in one time zone and people are living in different time zones.

Speaker A: Exactly.

Speaker C: So it's asynchronous kind of way that it cannot work that we are at the same time always present. So it's more like sending messages, sending emails, leaving notes and then agreeing on an ad hoc meeting. Okay, let's discuss this a little bit more. So it's very flexible way and it's a uh, community way of doing. We don't have any hard timetables because it's impossible and then we just drive forward what actually makes sense at the moment. And then there might be silence on some site for months or even a year and then they come back again. Okay, now we have this kind of thing.

Speaker A: Did you tap into communities that you

Speaker C: didn't know Existed some of these communities around the odp. ASPEC is where I actually went and uh, also collaborate with them. But then we had to create our own because there was no product community in this sense that what we were thinking because we didn't come from the data, we came from the business side. We had the difficulty that this kind of data product thinking and contract around it, they are very data and technical orientation and then you have these communities, technical developer communities and everything else ongoing. Our focus is on business level. How do I get the CEOs and CTOs and uh, these guys into some kind of discussion and they don't have this kind of certain channels for discussions, they go to conferences and then and things like that. It doesn't work for me. So it was about tapping into existing but then at the same time trying to figure out how do I create my own kind of global community.

Speaker A: Okay, cool. And I also see that when I check the latest version of your standards, this version 4.1m, I think at the moment you're actually showing a shift from pure technical specs to also including business strategy like KPIs or OQRs objective and key results. That makes sense. We need that in open data product specification. Why didn't we really realized that earlier? I mean do you see an impact on that? I mean it's always good to have a goal as well, but how do you make sure that the goal is not cluttering the technical spec and the uh, independence of the techniques?

Speaker C: Yeah, I think it was 4.1 when it actually, if I recall correct when I introduced it, even the 4.0 was kind of a turning point. Everything changed on the specification completely. It became uh, modular. You can combine, you can reference to see different parts of the specification. So the logic of the specification changed into version 4.0 and, and then because I started to think and see in discussions with different kind of companies that okay, there is a clear missing link between business level objectives and the data product. We don't have it. So business people cannot kind of see why we are building data product unless there is some kind of linking between the business level planning and then the actual data products that are managed somewhere. So that's where the I started to model. And that result, what we see is one of the five options that we actually kind of simulated and tested it and thought about, okay, how do we do it, how do we make it obvious? And then we brought the element of uh, product strategy. It's a combination, it's not a mandatory, but it's a combination of two things. It defines the business level KPI into which this data product is contributing. And then at the same time you have the product KPIs which are measuring the performance of the product itself. So it is serving the business and also the product manager that how is my data product actually performing at the moment? So, long story short, we needed to have a clear connection between the business and the product itself.

Speaker A: And I really think that is true for many of these technical specifications. And I love the thing that you actually include business and strategy more into the standards for an open data product because, uh, you want to be able to automate that connection as well. Business strategy, goals, vision of your board. What kind of data products might be interesting for us? I like that one. So let us go to the fact that you're from Finland. We said that in the introduction. Uh, you're from Tampere. Tampere. How do you pronounce that?

Speaker C: Tampere.

Speaker A: Tampere. Tampere.

Speaker C: Yeah, it's the same.

Speaker A: My Finnish is not what it used to be. Jarko. Let's go to music. When my first child was born, I was so happy that I needed some outlets. I went to a concert of a Finnish band, the Leningrad Cowboys. And they had this big clown shoes and nice dresses and huge hair. And they were making very optimistic, enormous, energetic music. I loved it. But I also know, like you said, there is this V? Taki music, the sad melancholic music in Finland. And when you moved to Abu Dhabi, you were exposed to the Middle east music, which is a lot about love and romance and so on. So how was that transferred to you? Did it do something with your music taste?

Speaker C: Yeah, the melancholic music mentioned as an example, I always had the taste for some kind of hard rock or some even metal or something else. And still the lyrics and everything is always kind of, uh, important. But the tone of everything is very sad. Everything goes bad. Your life is going to be a misery and your girlfriend is going to leave you. You're going to lose your house and everything is going to burn down. So that's the typical story, something happening. Um, but still the lyrics are telling a story, but you need to think what it actually says. There's a hidden meaning always. It's not like repeating two or three words on the song and then that's it. No, there's a story you don't see from the lyric. Yeah, you need to think and everyone is going to interpret. What's the lesson learned, for example, or what's Actually the meaning over there and it can be very hard to find. So that kind of a, uh, Melancholic. Yes. I don't know why, but it kind of uh, it's part of me that I love it and I need it. It's my lifeline. I need to listen it even in Abu Dhabi a lot just to kind of uh, relax my, myself. So for some reason this kind of heart and everything relaxes me a lot. But then the Abu Dhabi as you said, is. The Arabic music is very different. And uh, my cultural journey in this Arabic culture is still in progress, we can say. So I haven't yet adapted anything and I stick with the Finnish music and the old songs and uh, kind of something familiar.

Speaker A: So you definitely finish. I see you driving through the Arabic desert with this Finnish music, probably even in a Swedish car. I like that one. But there is another thing. One of my favorite writers is actually Finnish Arto Pasilina. And for the listeners to the podcast, if you haven't read Arto uh Pasalina, you're lacking something. So you should really read at least the Year of the Hare, uh, which is his, I think his most famous classic, where a man hits a hare with his car in the woods of Finland and gets out of the car and is starting to look for that hare, disappears in the woods and then he actually quits his cynical city life and wanders through the Finnish wilderness to find meaning. And while doing that he just comes into the most weird and surrealistic circumstances and he just treat them as normal and he just goes on and goes on and goes on. He just stuck. And I love that kind of books. I think it fits very well what you're describing here. So it represents the Finnish culture just as the music and these things fit together. While you're digging your ODPS tunnel with your bare hands. Jarkone, uh, finished as you are, uh, you're convinced it will work? Well, we see you're using a shuffle now too. So is that still sisu? I think you're starting to use generative AI in general, automation. I mean, that's no digging with your hands anymore, is it?

Speaker C: No, I agree. I think I need to clarify it a little bit because some people who don't know it, they kind of easily make a judgment call that it's a stubbornness. It's not about stubbornness, it's uh, a resilience and being very goal oriented and not giving up, though giving up is not easily an option. You push it through eventually, but how you get there is something that you are wise, you're tactical, you're looking options and if it doesn't work today, sleep on it and try again tomorrow. So continue as the story in the art of Linna book. So just go on and actually go and that's it. So accept what comes, adapt to it and then go forward. So that's kind of thing.

Speaker A: I love the YAML, uh, automation thing. I mean I've been following UML standards, the RDFL standards, the specifications for the XML DTD stuff and now we got the JAML uh specification JSON we have. So there's a lot of that kind of frameworks and I think generative AI is very good in helping you automate that and actually improve that kind of maintenance.

Speaker C: I agree. And the ODPS standard was until 3.0 it was JSON and then we switched to YAML. The YAML was already very familiar to me previously from CI, CD pipelines and automation. And it good for machines to read, it has enough structure but then it's for the human also very easy to understand and read and follow. Much easier than as you mentioned, XML and all these things that I'm also familiar with. I hated it to my guts when I saw it and I had to use it. I love the lightness and easiness of JSON and now the YAML. But yes, the YAML was picked because of practical reasons. It's as said, easier to understand, easier to process developers and automation builders and tool builders are familiar with YAML already and they like it enough. And then we found out that the sibling standards, for example data contract standard is actually using also YAML. So we were kind of aligning on this data product level with different kind of standards that we are at least using the same kind of approach on how do we define everything? And as you said, the AI has come to everyone's work. It means also my work in the ODPs. Definitely I use the AI even to model and mock certain things on the specification faster now. Um, but the thing is that I don't go to Claude or any other AI, give me an idea what to do. That's not the thing. So I do the thinking, I do the kind of initial systemic thinking. If this, something like this is done, what is the impact? Or here is this a good direction? Is this now aligned? So kind of designing everything before I even go to the AI. AI is helping me to make it happen faster, but it's not doing my thing. I'm in control. I decide AI helps me to actually get the results faster. It's a big help, I admit. It's part of my ODPS spec maintainer. Every day I use the AI on top of it. So. And sometimes I'm, I'm not happy with it and I'm yelling to the AI on my throne that okay, why are you doing this?

Speaker A: So what kind of feedback do you get? Finnish music?

Speaker B: Have you also noticed? And when we talk about culture and then also the aspect of now working more and more with AI, was there anything specific in terms of working together using an AI? Maybe even the mindset that is very different from maybe the Finnish Zizu talking now to the. Or thinking about the Arab culture. Is there anything that was striking to you?

Speaker C: It's not obvious, but I gave it a little thought previously a bit more that the Finnish mentality, it fits into the AI and automation world. As you know, as everyone might know. It's very precise and very accurate. The Finnish mentality is even very. We love gadgets, we love technology. So our mindset and brains is already tuned to this kind of reprogrammatic thinking. And uh, like the AI is behaving very predictability and everything else. So that was the mindset coming. And then in Abu Dhabi and in uae the mentality is more on speed and ambition. So the ambition um level here is much higher than even in the AI as said Abu Dhabi to be the first in the world AI native government in two years. So those sound like how can you even set that kind of goal and how can we even do it? But I love it. It's a good challenge. And now kind of combining this Finnish precise uh, working ways and then also not giving up and then this kind of good high level ambition, something world class level. So I kind of love the combination that I have here on that. But there is a difference on that kind of uh, ways of working also with AI. But these guys in Abu Dhabi seem to be more straightforward. Even go to the big goal, no explanations. Yarko, handle it, make it happen.

Speaker B: Okay,

Speaker A: Very nice. This is a very nice uh, thing. Okay. We also know that um, you went from Vitake Finnish uh, music set melancholic to the heat of the desert with Arabic music. And now your next destination might be Hanoi with the Vina house. Can you just tell us in two sentences what Wiener House is? We've never heard about it.

Speaker C: I don't think um, most people, probably Asian people know about the Wiener House music style. But yeah, it looks like that my next destination is going to be in the Asia and Vietnam, Hanoi or some province around Hanoi. Because of my personal relationship I have a fiance who's actually Vietnamese. So that's why we kind of planning that. Okay. Over time when we get uh, bored in Abu Dhabi, we actually relocate back to Vietnam and then decide what we're gonna do. Vinha House is some kind of uh, a dance music forum. They are very much alike all these songs but easy to dance. They use the Vietnamese language and it's very pleasant to listen. So it's far away from kind of a V. Kate or hard rock or anything else. It's very the opposite and they love it. They also have the same thing in these Wiener House songs. There's always a story, but not as sad as the phoenix present. These are sometimes more often happier. But of course their failed relationships is always uh, sometimes the story over there but not so deep but kind of this is the something similarity that I found that the story is there and interpret it now and think about it what it actually means. But it's different world.

Speaker B: When we talk about these different cultures and you moving to Hanoi, you've explained experience now Abu Dhabi, you come from the Zizu mentality. Um, what do you think about the future of global data spaces? When we think about us as global citizens sharing data, having open specifications, what is the future of open data spaces or in general data spaces? And what can we share in the future? How do we work together with AI approaching and developing so fast?

Speaker C: M. Good question. And this is always something that you are just guessing and trying to predict. Uh, but how I see that we are going to move away from data as a service thinking or that kind of uh. A very technical. As long as the pipelines work, as long as the data is flowing, as long as the APIs are up and responding, we're going to go towards the product thinking more and more and that the difference is that the product itself is built for a purpose. So you don't start from the data, you start from the opposite. You flip it around and then you have a purpose. Why you're actually even defining some kind of schema of a data product and why do you need the pipeline? So the starting point is different. And now I see that this kind of shift from raw data, ah, sharing well raw or a little bit processed. But anyway sharing of data uh, is going towards controlled data product which means that organizations will not expose everything. They will expose specific data product for specific use case with the clear rules. So trust and control actually define the future more than the data itself. So I see this product coming more than just fancy slides or speeches in a conference. But business level is still having a little difficulty to adapt to this one. They need a little help, more repetition, what it actually means, and they need examples. Like I said, the ODPS is a standard and you don't see the benefits until someone builds something on top of it. That's why we built the platform that we have that actually makes it very clear even for the business level, why do we have the data products and so on. But it's going to go there in my thinking, hopefully in a five year scale, every data set which is very data oriented becomes a data product. So we have data, but we're not anymore talking about data sets, but product on top of it, which are kind of abused. The masters of data, each of them has an owner purpose, miserable outcome. Uh, so it's a business oriented, not something that we're happy that it's flowing.

Speaker A: And if we use the ODPs correctly, then in five years we can also automatically compose our data products into our business promo. So the promise is there, the technology is there, the standards are there. Jarko, you did a great job in explaining it to us and uh, combining it with your personal story, which we love. Thank you very much for that. We'll learn some Finnish words down the way. We have one question left for you, Jaeko, and that is actually in Norway we have a very specific concept. Every family in Norway has their own. Nearly every family in Norway has their own cabin. And in this cabin there is a book that if you borrow the cabin or you're at the cabin as a guest, you're supposed to write into that. So we have our AI and data, uh, hutte book, it's called the Norwegian. And in that we want you to write your question for the next guest. What would that be?

Speaker C: Yeah, that would be, by the way, that Hunterrasciria. Don't even try.

Speaker B: We are not going to try.

Speaker C: So what I would actually write into that kind of a book, it's a question rather than statement that I would actually put there in that book. So what business outcome are you trying to achieve with your DAPA and how you can approve it today? So make the business people now think and come from the start leading the actual value creation from the dapa. So inspire them to take now the leap, be bold enough, have SISU and go forward.

Speaker A: So what business outcome do you want to achieve with your data and how

Speaker C: you can approve it today?

Speaker A: Yes, exactly. And that is what we will bring to our next guest. That can open the Hoode book next time. Thank you very much, Jarko. And, um, thank you for all your sharing, all your knowledge.

Speaker C: It was my pleasure to be here. Thank you.

Speaker B: Thank you so much. Have a lovely day and then talk to you soon.

Speaker C: Bye.

Speaker A: Bye. Bye, bye, bye.