Utopias, Dystopias and Today's Technology
Utopias, Dystopias and Today's Technology
Coding Nepal's Future: A Unique Blend of High-Tech and Tradition in the Himalayas
Venture with us into the unique conversation nexus of Utopias, Dystopias, and Today's Technology. This episode features the thought leader, Umesh Upadhyaya, Co-founder of HPCNepal, who, along with our host, Johannes Castner, navigates the intersections of high-performance computing (HPC) and the evolving landscape of Nepal.
Unfurling Umesh's professional chronicle from the international corridors of UNDP and the International Seabed Authority to fostering an HPC revolution in Nepal, this discussion illustrates the comparative dynamics of cloud and HPC, primarily highlighting the latter's quintessential necessity for Nepal. Uncover the challenges cloud platforms pose in Nepal and the empowerment brought about by global supercomputing conference learnings.
Witness the amalgamation of high-performance computing with low-tech devices, foreseeing an educational and research revolution in Nepal. This episode also illuminates the practical utility of supercomputing in weather prediction and the influence of Moore's Law on HPC evolution.
Stay tuned as we elucidate the prospective role of IoT in HPC, Umesh's experiences with global organizations, and the pivotal role of supercomputers in nurturing computational research ambitions in young minds. Our dialogue further encompasses the unique geographical challenges and opportunities for HPC in Nepal, international collaborations, and the essence of digital inclusion.
Concluding with the vision of mobile-enabled HPC, the incorporation of collective intelligence in contemporary scenarios, and harmonizing traditional wisdom with advanced technology, this episode serves as a trailblazer for future advancements. We also shed light on the burgeoning influence of technology in Nepal, demographic and economic trajectories, green energy avenues, and the ethical quandaries surrounding AI.
Our conversations are not mere technological showcases but serve as windows into the societal implications, the practical workings, and the future trajectory of technology. Connect with us and participate in this lively discourse to reimagine the future.
If our content resonates with you, show us your support by liking, sharing, and subscribing to our channel. Stay connected, stay updated, and together let's envision a technologically advanced, ethical future.
Hello and welcome. My name is Johannes and I'm the host of the show. Today I'm here with Umishe and we will be talking about high performance computing in Nepal. Um, this is a fascinating topic. I think I, I am always interested in how super modern, um, technologies are implemented around the world, Theme Song (written, performed and mixed by Neal Rosenfeld, sang by Jennifer Youngs) Venture with us into the unique conversation nexus of Utopias, Dystopias, and Today's Technology. This episode features the innovative thought leader, Umesh Upadhyaya, Co-founder of HPCNepal, who, along with our host, Johannes Castner, navigates the intriguing intersections of high-performance computing (HPC) and the evolving landscape of Nepal. Unfurling Umesh's professional chronicle from the international corridors of UNDP and the International Seabed Authority to fostering an HPC revolution in Nepal, this discussion illustrates the comparative dynamics of cloud and HPC, primarily highlighting the latter's quintessential necessity for Nepal. Uncover the intricate challenges cloud platforms pose in Nepal and the empowering transformation brought about by global supercomputing conference learnings. Witness the intriguing amalgamation of high-performance computing with low-tech devices, foreseeing an educational and research revolution in Nepal. This episode also illuminates the practical utility of supercomputing in weather prediction and the significant influence of Moore's Law on HPC evolution. Stay tuned as we elucidate the prospective role of IoT in HPC, Umesh's enriching experiences with global organizations, and the pivotal role of supercomputers in nurturing computational research ambitions in young minds. Our dialogue further encompasses the unique geographical challenges and opportunities for HPC in Nepal, international collaborations, and the essence of digital inclusion. Concluding with the vision of mobile-enabled HPC, the incorporation of collective intelligence in contemporary scenarios, and harmonizing traditional wisdom with advanced technology, this episode serves as a trailblazer for future advancements. We also shed light on the changing food nutrition landscape, the burgeoning influence of technology in Nepal, demographic and economic trajectories, green energy avenues, and the ethical quandaries surrounding AI. Our conversations are not mere technological showcases but serve as windows into the societal implications, the practical workings, and the future trajectory of technology. Connect with us and participate in this lively discourse to reimagine the future. Don't forget to tune in to our next episode revolving around AI and music. If our content resonates with you, show us your support by liking, sharing, and subscribing to our channel. Stay connected, stay updated, and together let's envision a technologically advanced, ethical future. Theme Song (written, performed and mixed by Neal Rosenfeld, sang by Jennifer Youngs) and, uh, this is what this will be about. So let me, uh, introduce you to Umesh. Umesh, uh, is a co-founder of HPCNepal, um, and he has a bachelor's degree in electronics and communication engineering, as well as an MBA with specialization in management, science and systems. He has experience working with high performance computing systems at International Center for Integrated Mountain Development, where his involvement was with Atmosphere Initiative and SERVIR-NASA projects. Mr. Upadhyaya also collects experience working with International Seabed Authority, U N D P and Elections Commission in Nepal. Mr. Upadhyaya is a recipient of a full grant to attend international Supercomputing conferences, in 2017 in Germany. And, um, he was a first recipient of Inclusivity grant to attend Supercomputing conference, uh, in, uh, also in 2017 in the United States. Great. Let us get right into the conversation. Um, hello Umesh, it is great to see you. Welcome to the show. And, uh, let me just start right away with a question, um, that, uh, you know, how how did you start, um, you know, the, the HPC in in, uh, Nepal. How did that come about and what is the process basically that you went through to install the first high performance computing cluster or computing system? I am not even sure if it's a cluster, um, in Nepal.
Umesh Upadhyaya:Hi Johannes. Thank you so much for, um, lot of information that you've collected about me. Um, yes. Uh, So, uh, high performance computing, um, was an exciting arena for me when I started my, um, mid of my IT career. Uh, and then, um, I figured it out. The scientists and the researchers, they do use, um, big models, um, uh, in, uh, using supercomputers. And then, uh, there was this small cluster, uh, that I worked in, uh, in one of the organizations and they, and the scientists were always, uh, were more focused in research areas related to weather forecast, environment modeling and sort of things. So I worked there as a system administrator, um, you know, providing support to, uh, The, uh, the researchers, uh, the domen, uh, people who used, uh, those super computers, it wasn't a big, um, as we talk about the industry, the big industry, it was very small. A few number of clusters. Uh, probably 10. I, I think it's, it was 10. Um, then sort of working on the,
Johannes Castner:where was this?, where was that located?
Umesh Upadhyaya:This was in Nepal
Johannes Castner:itself. Okay. Yeah. So they already existed. So, so they already had some high performance computer clusters. Was it at the university?
Umesh Upadhyaya:Uh, no, it was in one of the organization. It's, uh, it was International Center for Integrated Mountain Development. So yeah, that was the organization where I worked and they had a small, um, cluster. I see. And so that was, uh, where I first knew about supercomputers and how they functioned. And then, um, after working for a few years there and then going abroad and then coming back, uh, you know, uh, there was also a news that came, like, Nepal is going to have its own supercomputing facility at the university level. Um, so CERN um, as you know, they, um, donated, uh, lot of hardwares to university. And then I had, uh, I had, I had gotten an opportunity to volunteer at Kalifornia University to install some softwares to, you know, configure those hardwares. And it was pretty much exciting, uh, time for me, um, while doing that. And, uh, prior to that also, I had envisioned like, maybe it's a, it's a great idea to start, um, a project, uh, relating to supercomputers because, uh, many nese don't know the fact what a supercomputer is and, uh, how to get started over it. And then I had, um, international community, international visits that I made during that period and that actually gave me a lot of insights on how supercomputer in other world started. And then I, uh, I formed this non-profit organization called HPC Nepal. And since then we are working, uh, not to a larger scale. We are hoping some bigger, um, things are coming up on our way, but yes, we're working, um, into some domains. Uh, in the high preference competing, uh, part.
Johannes Castner:Yeah, that's, this is great. I, I have a question there. So it says on your website that CERN also invested in you or, or gave you hardware, or how did that work and how did that come about?
Umesh Upadhyaya:So I cannot, uh, say pretty much on the overall process, but I, uh, I got that news. We did not receive that hardware. It was the university, university at Nepal, which received the hardware. So we do not have, as an HPC Nepal, we do not have our own hardware systems. Uh, but yes, uh, we do, uh, we do some research on our own. Um, and then we use our own, you know, small environment. Where we sort of train our people and then do some, uh, pretty much, uh, basic works on, uh, supercomputer. But we are yet to, uh, get some, uh, more news coming up, uh, on how we will be functioning more, uh, on probably this year or next year on user, uh, super computers.
Johannes Castner:How long have you been around again? What, what was your, uh, founding year? When did you find, uh, uh, h HPC in Nepal?
Umesh Upadhyaya:So I think, uh, it has been four years now.
Johannes Castner:So, so what, what is a super, so, okay, you mentioned also the word super computer, which is a bit different from high performance computing. I'm assuming this, uh, related terms, I'm assuming, could you explain, uh, Maybe to the audience and to myself also. What, what is the difference between these two terms? Are, are they interchangeably used or do you use them interchangeably or is there a distinction? And then also what, what can you not do with, uh, without it? Is there, um, so basically, what are the use cases that are specific to doing high performance computing or super computing?
Umesh Upadhyaya:Okay, so. Um, as, as from my understanding, uh, what, um, what, uh, the matter of fact is like the hyper performance computing is any, any computers that, uh, um, you know, uses a massive sets of data sets, uh, unlike the, um, use of laptops or any other, uh, such desktop devices that we can use such big datas and then manipulate to get that output is where we call that, uh, uh, high performance computing system. These terms are quite interchangeably used in the industry, and, uh, it's, uh, for me it's very, uh, much of, uh, you know, of, uh, context where I use supercomputing and high performance competing mostly, uh, you know, quite often related to the matter of fact that how we use it and how I, how the industry explains it more, uh, more of that kind. But, um, uh, to that extent where. Probably how we use high performance computing or supercomputing uh, words. Um, actually, uh, I have not gone to the depth of that matter on the difference, uh, between that. Um, so yeah, just like I said, I use that, uh, quite interchangeably on that. So, um, for HC Nepal, one of the use case, uh, that we are using currently is the, uh, use of high performance computing for weather research and forecast. Um, the, uh, universities here and research organizations, they do use, um, h HPC systems in Nepal for weather forecast. There isn't big industries as such, but of course there are a lot of use cases that we can work on here. Mm. Uh, uh, currently what we are doing is, um, how to commercially viable, uh, commercially use the old, uh, phones, mobile phones, Android, and the. Uh, iPhones, how we could leverage their CPUs or, um, you know, GPUs, um, and then maybe use them as a high performance computing, uh, part or how we could use that. Um, of course there are few research that has been done and the recent one was also conducted and there one of the universities in Russia actually has already conducted and it's, uh, very, uh, successful as well. Uh, it has already been, been done. If you look into Boeing systems, uh, zeti at home or sort of other things, they already have done such volunteering, computing for fraud forecasting and many others. So we are trying to do that. But yeah, we are still progressing and then we are still in the, in the visibility part where we could do that, uh, or not. So that's the current situation where, um, we are working, especially with the use of, uh, hyper computing.
Johannes Castner:So it does it, how does it compare with working in the cloud, for example? Is it, is the, for example, is the Google cloud not available or the AWS cloud not available in Nepal? Because these seem to be also high performance systems, right? Because you, I, I can do massive computations on those clouds. That's how I, I guess, been using high performance computing. So I wonder is, is that available in Nepal or is that not available in, is that the issue or, yeah. Could you, could you speak on this comparison between cloud and high performance a little bit.
Umesh Upadhyaya:Yes. So, um, so in Nepal, like, um, the high performance computing cloud, especially the facilities, um, they have not been used by much, uh, industries or, you know, research organization. It's, it's available. You can use Google Cloud and aws. But, um, the fact that, uh, Nepal has to have its own homegrown facility, one of the facility for h HPC is actually a must needed part because we are, we are in the, in the domain sector, if you look into, there are so many sectors like hydro, hydro, electricity, Nepal is very prone, prone to earthquakes. Uh, so, uh, that homegrown industry is itself needed. And of course, uh, on the cloud level, uh, if you look into it, um, that might come as, uh, Uh, more of a expense, uh, to the industry here, um, because the research domain hasn't scaled up to that part where we could use the cloud features. Uh, so I think starting from small clusters and going forth on that area would be better. But yes, um, uh, use of cloud is limited to non-HPC sectors, but it hasn't been used, uh, to the s spc part
Johannes Castner:And and why is that, again, I'm sorry. I don't understand why, why is the cloud not applicable or what cannot not be used for high performance computing, just across the board? Why, why doesn't need its own one. What's the advantage of having your own?
Umesh Upadhyaya:Yeah. I mean, like, there are a lot of advantages of having, uh, uh, uh, own super computers, a as such. And then there are also reasons behind the, um, uh, you know, cross data, uh, uh, uh, the, uh, cross border, uh, datas that we might not be, uh, able to, uh, be applicable to cloud, uh, platforms. We could, uh, be using our, the data in, uh, the homegrown supercomputers, uh, Also the fact that many in Nepalese people, or many technology, or many researchers, yes, do not know how to use the cloud. The cloud industry is not so large enough that they use, um, uh, and then the frequent trainings might be needed to use that. And that would also add additional cost on, you know, bringing, uh, you using the cloud and others. Also, there are some, uh, there are some policies that does not let, um, the US dollars of the economy to go abroad. Uh, there are some policy restrictions, um, in Nepal. Uh, so that will not allow, that will, I mean, like that is a, uh, is a problem where, uh, you cannot pay as much as you want, uh, for the cloud. There are some restrictions with that. So that would be, um, one of the reasons as well.
Johannes Castner:That makes a lot of sense actually. So, so let me ask you a question about the, you know, on, on the website it says a lot about, um, uh, Nepal's participation in global computing supercomputing conferences. Um, what, what are the, the key takeaways from those and, and what, how they, how have they been applied thus far in Nepal already?
Umesh Upadhyaya:Okay, so, um, so, uh, uh, Nepal, um, as a matter of fact, um, Uh, if you, if you look into China and India, they have massive, uh, uh, use of supercomputer use cases, and they do have, uh, they're also growing very large in terms of, uh, um, exit scale or, or, you know, use of use cases, various use cases. They have multiple universities where they're using supercomputers and they have also been listed in top 500. And that sort of, you know, we, being in the middle of that, uh, technology, big technology designs actually, uh, there was a, there's a thought process that we also might need and then use it to us, our extent, to what extent we can go and learn from these, uh, countries and then create values on our own, on how we could, uh, you know, use supercomputers, uh, on that part. Uh, uh, so. Uh, I think, yeah, yeah. That, that's the, that's the most. And then the other, other, uh, part is like, there are so many international, uh, Nepalese, uh, there are Nepalese people, uh, researchers and scientists abroad doing, um, research, uh, and sort of, uh, their own, uh, work in high performance computing system. So they want to contribute to Nepal in some way. Where that could be more effective, um, where they could be, uh, you know, they could provide some, uh, volunteer support to community support or in research domains where they could be helpful. So I had, uh, connected with those kind of network as well and the international communities. Also are so, uh, helpful in, you know, providing trainings. And then there is also exceed and then resource, uh, where they try to provide to, uh, sub certain support services to the, uh, to the resource constraint environment. Um, like Nepal in terms of h hpc. And, uh, these communities ha have, uh, evolved, uh, so much and I have been connected with them. So I got, uh, uh, quite a few trainings as well on that part. Uh, in Nepal we are also training few people on hpc. And then, right now we are also doing some, uh, work on, uh, how we could use, uh, leverage the use of, uh, mobile devices for, uh, supercomputing, for use of HPC. So these things have actually happened, uh, with us. And uh, there is also o other international communities who want to tie up. Further, uh, with us, uh, give us some work to get started. And that's happening pretty much soon. Um, yeah, that, that's, that's, that, that has come as a possibility so far.
Johannes Castner:That's really sounds great. I, I had a guest on my show recently, a few weeks ago named, uh, Ananya Agrawal, and she was speaking, she, she actually is in India, but also in the Himalayas. She worked in the Himalayas in, in, uh, villages where she was delivering very innovative, um, Uh, learning solutions. We are very low tech devices. So these, these people didn't have even smartphones, but they had some kind of, you could say dumb phones, right? So they could call some numbers. And then they were sort of connected with this learning system that she describes and, uh, about, and, and the learning was all about opening businesses and starting things and, and developing and, and in innovating from materials even all the way. So I'm interested in this low tech solution as well, and how it might be co connected to high. High computing performance is in the cloud or some somewhere else, right? So, so maybe people have like these little phones and they could interact with a high performance computer. Is that something that's possible? And then how is it used for education versus research? You know, you speak a lot about research and so on, but what about, you know, all the way down to the. To the lowest level of education. Are there some applications of, of, uh, high performance computing that you can see, either how people learn, how it affects their learning or what they learn even?
Umesh Upadhyaya:Uh, yes. I mean, like, on the smaller scale we do have like Raspberry Pis and then, um, you know, uh, uh, you know, sort of those devices that can be used. And you, one of the universities in uk, uh, I think they have all, they also use, uh, those respiratory pipes to build a cluster and, uh, in. In, um, it's great always to see in supercomputing conferences also that people have been using those small devices and then readily used it for a lot of, uh, things including, uh, one of the, one of the areas always, uh, like I'm mentioning is the weather forecast because that's the domain where Nepal, uh, is using it and I'm more fascinating with it. Um, having said that, yes, like the small smartphone devices can be used, but we've not, uh, we've not pretty much, uh, aware on whether that could, that phone could be, could be, those phone could be used for parallel processing. Uh, part rather they would use the distributed competing, which is different from, uh, what high performance computing, uh, normally do. Uh, but as a matter of fact, yes, um, those phones could be also used, but that's, uh, that, that has, uh, to be, you know, uh, we are still, uh, Researching, doing research on that. And there are international communities also supporting on that if it's a possible, there are a few other, uh, uh, universities, uh, that we are in communication with. And, um, if that's a, that's a possibility, that's a another part. But yeah, on the educational small, uh, uh, K12 students or uh, students level, what we could do is, you know, use the laptop itself. There are container existing systems, right, like the docker and all sort of things that can be used even with, uh, high performance computing. So we can bring that whole data center to a small laptops, you know, collection of laptops or the lap, the laptops itself. We could do it. Um, the other ICE mentioned already, like the Raspberry Pis and then phone systems could also be used. Yes, there are, uh, these are the various, uh, ways we could work on it. And I think we have set a few plans. Um, I can only tell you, uh, what, uh, what happens like maybe. Within a year, what we will be doing on that part. So yeah, that's it. Great.
Johannes Castner:Yeah, I gotta have you back for that then when, when you have, uh, when you're further along your journey. But, so, um, let me ask you, uh, this, this is all, uh, you know, could you get a little bit more into the details? It's like as to, for example, how do you use super computing for, so for, for example, weather forecasting. How does that actually work? So just, uh, just so that I, I and the listeners at home have a bit of a sense of, of how that works in practice.
Umesh Upadhyaya:Okay, so, um, so first, um, you have a few, uh, clusters of hardware. Um, then you set up that using any open source tools like operating system, like the ready available ones you could use center, which probably, um, not anymore, but you could use Rocky Linux or any sort of, you know, freely available operating systems. And then on top of that, there are certain HPC tools that you can install. Um, and then you can install the weather forecast application on that part. And, uh, Uh, on the visual lesson part, you could work on how to get that visual lesson part for the data part. Uh, you'll readily get those data available in NASA websites and other satellite data. You can, uh, readily get those, so you can get that data readily from, uh, websites. Lot of, uh, you know, available sites and then you can use that, install the application and, um, get the data that you want for the weather forecast for that domain. Knowledge. For weather experts, we need, we might need weather experts, weather forecast experts who are more, um, into that, um, you know, uh, um, you know, more, more, more, have a more profound knowledge on the weather systems, how it works. But on the installation and configuration part, it's mostly the, um, system administration parts, which can collaboratively working on it. We could. Uh, and we could, uh, we could, um, as a matter of fact, so, um, we could also select, uh, a domain area for like, for Nepal or for mountainous region. And then we can, uh, uh, you know, uh, do some predictions over a period of time. Like, um, six kilometers, 12 kilometers or, and then the time range of that, and then, uh, run, uh, forecast, uh, on that.
Johannes Castner:So is there any AI involved in this at all?
Umesh Upadhyaya:Mm, yeah. The industry actually, uh, gives a lot of, lot insights into the AI and ML part, but we haven't, uh, I'm not much aware. Uh, and then we haven't used that so far. For our context.
Johannes Castner:I had my, my. Yeah, my, my long term mentor, uh, uh, Neville Newey on last week, and, um, we were talking a bit about the evolution of artificial intelligence in computing, and he was, you know, addressing this very famous, very well known rule of computing, which is called Moore's Law, you know, where, where the, the computing power, um, doubles every so often and, and it go, and it, it really kind of is an exponentially growing, um, uh, you know, factor this computing power has, has that affected your work at all? Is this like something you have sensed and seen within your span of working, uh, with, uh, high performance computing? How does it affect high, high performance computing in general, or supercomputers? Could you, could you speak on that?
Umesh Upadhyaya:Um, um, I think, uh, In, uh, as per my knowledge, what I think is like, um, since Moore's law is pretty much, uh, not very practical now, uh, and then we are moving, uh, that curve, uh, pretty much higher than what we actually learned, uh, um, previously. Um, what I see is like everything including ai, machine learning, deep learning, whatever it is, are all conve conversing to the high perform computing system. Or maybe they will get conversed to, um, other kind of, uh, you know, uh, HP systems or maybe, uh, for future systems that we do not know what we call right now talking about quantum and all sort of things. But I think the convergence is all there. Um, while that convergence happens, um, I think the use of, um, CPUs and the use of GPUs. And when we are talking about petaflops and to that level of large scale data and all sort of, uh, things, um, I think this is going to be very much, uh, uh, a game business, uh, where people sort of the industry in a competition mode on how to get into bigger, uh, on a larger, uh, you know, uh, domains of, uh, work and how to, you know, improve the processing capacities and all. But for, uh, sectors like, um, like, uh, well, like, uh, small and medium enterprises or other part, um, there are possibilities where we could explore the grid computing, for example. So the grid computing would mean we could use the, uh, use the. Servers, uh, old devices, mobile devices, servers, laptops, tablets, or any desktop competitors for that matter, anything. And then come with an, uh, uh, um, an architecture, uh, which would leverage the need of users at all levels. Uh, for example, the big supercomputers could also be used, the lower scale. Uh, you know, raspberry pies could be used, the laptop phones could be used. And then the, so I think that harnessing of that, uh, green computing would also equally be coming up in future. And that would add more to the use of super competitors in use, in how we could use in the, um, low scale, uh, uh, devices to, uh, to the higher, higher level of devices that you currently, we are using.
Johannes Castner:This is kind of related to the Internet of things in some ways, right? How, how does that factor in there? Is, is there a connection to that, to that area?
Umesh Upadhyaya:Um, what I think is like, um, With iot, we will be getting lot of informations and data. And ultimately when we are doing a lot of calculations, um, some, some might need very memory intensive calculations, some might not need that. So based on that parameters, yeah, I mean, like we iot devices, um, yeah, I mean like these, these kind of devices will be used to generate the data that's actually used for, uh, that we will be using with lot of tools. Uh, and then using the supercomputer more and then simulation models, we will come out with output, um, maybe on a bigger scale, on a larger scale that we have been doing. And that will also generate, uh, bigger capabilities on how we might further go on implementing supercomputers. And I think that's, that's what will drive the industry. And then the current scenario further, Yeah, I mean the IOT devices, uh, as such, like you said, uh, the mobile devices could be used for data collection mostly.
Johannes Castner:So let me then ask you, um, a, a bit about something different. Um, could you tell me a bit about, um, your experience working with government organizations and UN agencies? Has that influenced your work a great deal at h HPC Napal?
Umesh Upadhyaya:Mm uh, right now, Not very much. Um, yes, we, uh, with the, with the UN agencies and then, um, with the government sector that I worked, uh, it was mostly on the IT support part rather than on the h hpc uh, part. Um, um, I did not work there with the HPC. And yes, we are somehow tying up with government, uh, organizations to install and then configure their system. Probably train them further on how to use the h HPC systems. But, uh, yeah, I, I think that that's pretty much it for now. We haven't scaled on, uh, further, but yes, on a university level, we have done some, some, uh, um, as HPCNepal has done some, uh, understanding, we have done some collaborations where we will be, uh, guiding the, uh, supporting the university on, uh, on, uh, installing and configuration of supercomputing, providing them trainings when applicable. So sort those sort of things are there.
Johannes Castner:So this is, this is part of your bigger mission also. Right? And also there is a, you know, on your, on LinkedIn, it also, you, you are talking about the enormous potential, um, that computational research can bring to young people and to young brains of society as you put it. Um, c could you tell me a bit about that? Is there already something you're working on, or is this just a part of your bigger vision and how do you connect your, what, what you're doing now to this vision? Could, could you speak on that a bit?
Umesh Upadhyaya:Yes, sure. So what we are doing currently is, um, not much to a bigger scale at the university level. Yes. Few, few people on the university level. We, we train them and we have done some research. Uh, uh, so yeah, I mean like, uh, so, uh, the fact, the truth is like, uh, many universities or institutes in Nepal do not know. Much on the use of high performance competing systems or, uh, high performance competing system itself. And what we thought was like, um, the, the hyper, the, the, uh, the undergraduates or the graduates at, uh, technical colleges, um, they will have to produce some master level thesis or dissertations for that matter, and. We have a plan to go to them and approach them on the use of, uh, supercomputers, and then probably they could come up with an idea and then we, you know, sort of gather those people on that similar domains and then help them on the scientific part so that they do, uh, they do the part on the technology front and then come up with something that will help the industry, that will help the university. And it'll also groom up that, um, uh, s HPC level, uh, you know, the education level quality as well. But since, uh, like I said, like the industry is not, um, ready for it because the industry does not cater to the HPC thing here. So probably that will be in, as in, uh, Uh, because many Nepalese, after undergrad or graduate, they travel, they go abroad for work or for further studies. So that might be the case. But that, uh, having said that, that would be a larger, uh, region for us for right now because we do as, as, as, especially Nepal, we do not own our own super competing facility. But we are on that, uh, we are on that, um, discussion with one of the, I cannot disclose you right, currently, but we are still in the discussion. And once we have that, uh, small, uh, clusters or set up of our own, then we will, uh, come up with, uh, you know, we will have joint ventures with lot of universities institutes. And then, uh, further on the development of the program, how we can go about it.
Johannes Castner:Are there, uh, this is a, a little bit of a different question, uh, re not related to the last one really, but um, more related to Nepal and India and you know, these areas where there are a lot of mountains. Is there anything very specific? Are they unique challenges that you have to overcome because of the terrain that you're in? Or are there also benefits perhaps, of being up in the mountains?
Umesh Upadhyaya:I think, um, I, I see there are a lot of benefits, actually. Benefits in sense that, uh, uh, these high altitudes in terms of technology, if you're looking into, yeah, I mean the, the cooling facilities is already there. We don't have to cool much anymore. Um, uh, on the, on the non-technical part, yes, the development work might need, it would be a challenge to, you know, pull up wires, uh, to that place. Um, people, the livelihoods would be like the ecosystem, the livelihoods would be like really challenging on the, those parts. Uh, Those fields. If you look into, if you look into the, uh, energy sectors, this, these mountains actually bring a lot of hydro power. Um, so Nepal itself is very rich in hydro power. So in terms of energy, if you look into it. So a lot of data centers actually can come here, and I, I mean, and leverage that hydro electricity power, which is more cheaper in many ways than, uh, using the other sources of energy because it's completely a green energy and it comes from the mountains. And, and then the use case itself, you know, for the weather forecast, uh, part is also one of the domains because in Himalaya, many people go for treking and then there are, uh, terras areas where, um, pilots might have a challenge, uh, uh, for the flights and all. So I think, uh, then that's, that's more of a kind that's, uh, that's needed, uh, one of the areas. Yeah,
Johannes Castner:I understand. That makes a lot of sense. Also climate change probably. Right? So the longer run Yes, because it's different from weather really. It's, it's, the climate is sort of an aggregate weather, I guess.
Umesh Upadhyaya:On the climate modeling, I think, uh, it's, uh, climate research, I think it's mostly, yeah. Uh, rather than at a country scale, it's more of a regional scale that can be done. Uh, like the whole Himalaya Range is, uh, I mean the, the coverage that most countries take part in, like India, Bhutan, China, Nepal, and other countries as well. So at a regional scale, um, the climate, climate, you know, modeling as such. Um, I think that this has been going on and then, yeah, that, that's a great area to work on and that's much needed actually.
Johannes Castner:Oh yeah. And do, do you cooperate with other countries? Uh, easily Is, um, Nepal in good relations with India and, uh, China?
Umesh Upadhyaya:Mm. Yeah. I mean, based on, based on what we've seen so far. Yeah. Um, the, that's, that's the whole, um, story of it. Yes. Technologically yes. If you look into technology domains and all the sectors, uh, yeah, I think we are very much in a very, you know, uh, uh, geographically also very close, and then very much, uh, uh, knowledge sharing, uh, you know, uh, areas, uh, between, uh, between the, uh, countries.
Johannes Castner:Well, so, and, and in terms of, uh, digital inclusion, this is a concept, you know, when, when a lot of people, when a lot of people have, uh, no access to computing in general. Right. Is there a way that you can see how, for example, something like Ananya Agrawal's work on low tech computing or low tech, uh, work can give people access somehow to, you know, to, to systems that may not actually be on their little low tech device, but be somewhere else, right? So that they can be connected to them and be included, uh, even though maybe they don't have a lot of money or a lot of, uh, even maybe infrastructure that allows them to do computing directly where they are.
Umesh Upadhyaya:Okay. I think, yeah. Yeah. I mean, like, I haven't, uh, Heard about her journey. Uh, as such, I will have to do some research on what she has done, uh, more and then maybe I can talk about it, uh, if it's a possibility. But yes. Uh, on the small scale, uh, devices, like you said, yes, there could be small, uh, labs that might be working at schools, uh, at institutions, small institutions which provide the data. Which can feed, uh, pro, you know, collect data and then provide to, um, um, uh, um, a larger set of, um, cluster computing environment so that that computing can happen there. Um, yes, probably this set of iot devices or data sets can, uh, leverage a lot of data that, that might need in terms of, uh, use case, uh, relating to the, uh, socioeconomic aspects as well, because that's also a domain where, uh, s SPC can be used, or even for nons spc, any technologies for that matter. Uh, but, uh, as I see, yeah, low tech devices from preferably, I'm not an IOT expert, but yeah, for devices, uh, could be used, could be better used for data, data, you know, getting the information.
Johannes Castner:Well also, you know, Well, I'm also thinking, you know, for, for, for the minds themselves, right? So if you have humans in the loop in a way they can learn something. For example, is it possible, for example, for someone, do you think it would be possible for someone to learn something about high, uh, performance computing from a low tech device, you see if, if you attach them to some cluster, they could maybe explore how it works or learn something about it? Uh, do you, do you think that is possible?
Umesh Upadhyaya:That is pretty much possible actually. Okay. So, um, yeah, I mean, like, um, that is possible. Uh, as I see for, uh, based on what the technology offers. Um, that's what we, I, like I said, we are doing some research on, um, uh, on the. Um, on using that, uh, on using that for, uh, what kind of use case, uh, has to be identified. Uh, first because not use, every use case can be run, probably or not every applications can be run over their mobile devices. Yes. Uh, generic use case might be very much. Uh, and, and then we can probably train, uh, people on, on a lower level of scale on how that can be leveraged into, you know, used, uh, over mobile devices or IOT devices. And then, uh, they might have an insight on how oh, how the s HPC systems work or supercomputer work on a, uh, larger scale. Uh, just an insight would be a great to get started somewhere. Right? Yeah.
Johannes Castner:And so then, you know, in general, do you have like a sense of, um, you know, there's this concept now that goes around the world. It's called, uh, humans in the Loop, right? So it's basically not just computers that you can use, but you can use human labor or some human activities to process something bigger, right? So for example, even it, it even goes back, back to, to a very long time ago when, uh, the second World War was fought and the English were fighting against the Nazis and, uh, uh, in Normandy. So they, they would, uh, attack the beach and they would know exactly what the beach was like because there were vacationers that were there before that took millions of pictures of the beach where they were hanging out and where they were vacationing. So British. Vacationers before the war had taken photos of little sections of the beach, millions of them, and they had sent them into the bbc. And you could think of this as a calculation, right? So it's like to understand the beach, you have these people sending in pictures. So, so this is just a very old application, you know? Now we talk about collective intelligence and you know, there, there's many, many, many applications. Do, do you, do you find that in Nepal as well? Do you, do you have a sense of that where, where humans can be part of the computation in some way?
Umesh Upadhyaya:Um, I think, I think yes. Um, if you look into the collectivism or the collective intelligence, I think the whole network of humans are a part of it. Uh, it's just not nipple, but the whole world should be a part of it if you look into it. And, um, and, uh, What our society, our Eastern Society provides, um, here and then what, uh, the West provides. I think that that should be a merge between that, uh, and then come up with that information, which has, which gives us more insights or which can be used as more to generate more intelligence. Just not the data, just not the information, but that could harness more intelligence and then that could be used for the better of the humanity, I think possibly. Yes. Yes. In terms of, um, I, I, I would want to give an example on that part where mostly if we look into people here, um, We did not have weather forecast or, you know, in generally previous people a hundred years back or 200 years back, people did not know, uh, about weather forecast. But they, I have heard that they had insights that tomorrow it's going to rain or day after it's going to rain based on their observations of what's going on around. So I think those information comes from their, yeah, I think those informations comes from their. Uh, one of the insights and also on the work that they were doing, uh, every day on the fields, also on the animal's behavior and how that was happening. And maybe they pro uh, you know, uh, forecasted on the weather on what it would be in couple of days or, or maybe yeah. One day. Right. So I think, um, yeah, I mean, there are some sectors we could, uh, use, uh, those kind of observations that we have completely missed using nature's, uh, you know, using the natural part, just not the artificial intelligence, but the natural intelligence that's already there. Um, I think that, that that part, uh, is what makes, uh, the collective intelligence more, um, uh, exciting, right. So I think, yeah, humans can be a part, even in Nepal, if you look into, yeah.
Johannes Castner:I think there's also a flip side to this, right? So this modern, the fact that we do weather forecasting or even, you know, on your Google, like we in the west, I don't know if you do it Nepal, but we, we have Google Maps, right? And if we drive from one place to another place, we know exactly when we get there and so on, right? I think in some ways it might have reduced our own intelligence a little bit because we now rely on this intelligence, you know, this kind of artificial intelligence and we just rely on it and we turn off our own thinking in a way. Do you see this as a potential problem? And, and do you think that there's something we can do about this problem that we, uh, also.
Umesh Upadhyaya:Um, yes, I think, yeah, I think, uh, I, I think when we are talking completely on the artificial part, we've completely, we, we should not be missing the natural part of it because, because the artificial words, uh, word itself comes, um, completely from the opposite of what, uh, natural is, right? So I think, uh, So what, what nature provides, what, what, what brings us, um, we as a human beings or as an animals. I think those, those parts of things should not be missed when we are talking about artificial intelligence. We should not be just talking about programming. We should not be just talking about robots, um, as a matter of fact or, you know, software tools or anything. But we should also come into the ground level of that humanity where, uh, how the process of we being born, the culture we, we grew up on, and how our, uh, grandfathers or forefathers worked. What were their, uh, you know, needs and how that information. Today has impacted the world and what it could lead further, uh, for artificial. I think those kind of merge in those kind of things. If we can bring, uh, together, like the con that convergence, if we can bring, I think artificial intelligence can be of a, a big possibility for, for humanity in future. But if you look only on the aspects of the technology part, um, well, it might provide a lot of information, a lot of ins, insights, and maybe we could also struggle with what is true versus what is false. But, but yes, overall what technology today is providing is magnificent. If you look into it, if you stay home while staying at home, we are having everything, uh, in a gadget. Uh, but I think we should also look into the, the unlearning process of. Our previous people because they did not have much to learn, they just had, uh, uh, opportunity or quite a possibility to observe. So when that observation happens, what could we benefit or what?
Johannes Castner:Yeah, absolutely. I agree 100%. And I also think that maybe. Maybe you have an advantage. You see, what, what makes me think then, when you, when you say these things, it makes me think that maybe Nepal has an advantage over, say America. Because in the US we already so far into the artificial, you know, we've, we've started so early now, we're, we're, you know, we, we've already had many generations of, well, many generations, not really, but we, we, we've already, you know, gotten so accustomed to it and maybe this did not yet happen in Nepal. And you can learn something from this before it happens to you in a way. Uh, you know, that, um, to, to compliment the artificial with the natural in a better way than we have done. Because I feel like a lot of, you know, for example, Facebook, you know, we, we, they, they, they advertise to us and we spend our time on it, right? And we waste our time really on it. And, um, you know, those kinds of things, they, they actually make us a little bit dumber. There's this artificial intelligence and it's actually causing us to become, Dumber, you know, dumb us down really. So maybe people in Nepal and, and people in places that haven't had Facebook for as long as we had and haven't had the internet even as long as we had. Maybe they can look at us and say, maybe not exactly like that. Maybe something a little bit different. Maybe we can learn something from this. Uh, do you, do you think there is something to that? Uh, no,
Umesh Upadhyaya:actually we actually, maybe in internet we were a bit lacked, but I think on Facebook we were pretty much similar times. Uh, as far as my knowledge, uh, is there, uh, yeah. Yeah. I think, uh, I think one sector which needs to be looked at, uh, most, more profoundly is, um, what medicines, uh, bring about, like for everything we now pop up on medicine for that matter, like anything happens, but, uh, what. Easter, what we talked about was more often, uh, Ayurveda, the Ayurveda word itself, it means like how we can, it's a study of the age, the word itself does not talk anything about the disease. But right now, medicines are talking about how we can treat disease. You know, it's already the word itself. It's already coming from east to disease and how you can cure. But if you look into the, a better part, it, the word itself means how we can increase our age, which means we ha we are going to have a better health. How can we have a better health? So I think, uh, that's a domain where, uh, a lot of research can be done on plant-based, uh, you know, uh, medicines mostly. So that's where I think, uh, um, east can bring a lot of insights and maybe, you know, the books, the old books or, you know, a lot of knowledge base, uh, that we can bring. And then that can also be lever, uh, leveraged over h HPC systems, hyper computing systems on how we can use it for humanity to, um, increase their ease or, uh, benefit for their health.
Johannes Castner:That's a wonderful idea. So I really love this idea of, of, of combining traditional wisdom. With modern technologies and modern insights. I think that's a great, that's a really great thing in general, and I think that definitely in Nepal and India has a lot to offer in this regard. I also, I've, I've eaten Ayurvedic food in this fantastic, it's, uh, you know, traditional, uh, Ayurvedic food, even without any modern addition to it. Uh, and it is fantastic. It is really good. It feels good in the stomach. It feels, you know, you feel good after you eat it,
Umesh Upadhyaya:even the food, right? So I, I have like, uh, like for example, um, it's a fact that what our, uh, grandfathers ate, you know, the minerals, the amount of vitamins, the minerals that, uh, one orange had. So if we eat it, 10 oranges now. So we get the, we get actually the same kind of vitamins that was there only in one. So we have to eat 10 oranges for now, now. But, um, what we know is the fact is that, oh, the, the same orange is coming, but we do not have an idea and the amount of vitamins, the minerals that it carries, right? So those sort of inventions or research has not been done so far.
Johannes Castner:Do you, you see this as a use case for h
Umesh Upadhyaya:Yeah, I mean, like we should know if we are eating, uh, the right kind of, and just not the right kind of fruit. Fruit. Um, I think most fruits are the right kind, but the, the fruit that we eat actually also should have, um, a certain level of, uh, vitamins, a certain level of, you know, um, minerals and, and, and whatever. Uh, so maybe we could calculate based on the soil fertility or may based on the use of agricultural, uh, you know, the kind of type of soils, the type of fertilizers being used in the soil, and whether that fruit, the product, agri product that's coming up, uh, versus on the organic level, what is the differences? In terms of even the values of the minerals or the vitamins, and if it's, you know, if it's just the fruit with not no value at all, it's useless to eat. Uh, just saying the fried fruit makes you healthy, right? So there are a lot of factors that could be, uh, looked into
Johannes Castner:This actually points to a potential for a much greater scientific approach to how we eat and how we exercise. You could, you could really, you could really turn our entire life in, in a, in a sense, into some sort of laboratory, right? You could really figure out. What we should do at all times, you know, how long we should sit in the, in front of the computer, if maybe we should start standing after some time. And you, you could really extend what you just said to, to almost every aspect of our life, right? Where we can really have a good understanding of what are the things that causing us pain. For example, uh, you know, if I sit in front of a computer in a particular way, if I change it a little bit, if I raise the desk a little bit, maybe the pain will be lesser, right? So, and all of these things could be systematically studied, right? So right now it's all relies on our intuition or our feeling in our own self, or like tracking things we do and we don't do and how we feel. But this could be, you know, done also on a large number of people so that we really have a good sense of what works and what doesn't work in everyday life, or we haven't even really begun to crack this open. Is that, is that right? Do you see that as well?
Umesh Upadhyaya:Uh, yeah. I mean, yeah. I, I see that what you're saying, uh, actually makes a lot of sense. Um, Yeah, I mean, like we, we could go on that journey of, um, data itself. Um, but yeah, on, on the bigger aspect, what we need to look is how it can bring wellbeing to people. Um, because if it, anything that we do is bringing wellbeing to them, uh, whether it's working on the computer or, uh, if they're sleeping like four to five hours and, and even eight hours, uh, for that matter, yeah. If a person sleeps for eight hours, is a fact that has given, given to us. But it might, uh, as well be possible that, uh, a person sleeping for three to four hours is more healthier than a person, uh, sleeping for eight hours. But there could be other, other posi, other, other things, uh, you know, that could be, uh, looked at. Whether, what, what is making such differences and what is the need of that particular person? Because it, it varies. And maybe that could be enhanced in terms of we here practice those things as yoga, like, um, um, the certain exercise or yoga practices, that's more, um, and then meditation taking. I know, I know. Yeah. I mean, like, yeah. I mean, uh, you know, you could also,
Johannes Castner:you might be interested to hear, I don't know what it was, but one of those years I, I read in, in, um, business magazine that, uh, yoga studios were the fastest growing business in America. So, uh, that's, uh, you know, that's your influence, you know, that is the eastern influence on, on the us. So there, there is, I don't know if this is true every year or if it's always true, but it was at, at least once the case Yeah. That yoga studios were very fast growing. Uh, the fastest growing business in the us. So that's, yeah, I mean, that's an interesting fact. And we have it here in England. We have it in in Germany.
Umesh Upadhyaya:Yeah. I mean, I had read one article which said like, what is going to happen, uh, if AI takes over everything, every jobs and every, uh, vocation, uh, practices that has been doing. And then one of the articles said that, well, if people have, uh, if everything is taken care, uh, on their work or studies, then the o only thing they could do is sit and practice meditations or do some yoga practices because everything, when everything has been figured out, it's time that they do do such practices. So yeah, I mean, like, it was pretty interesting on that one.
Johannes Castner:And music, I think. But that's, uh, you know, music and art. I think, you know, I, I don't think that music and art should be done by ai. It doesn't make any sense. It doesn't make any sense. But do you think it's not, it's happening already right? Yeah, yeah, yeah. But I can't listen to, it doesn't make so far
Umesh Upadhyaya:Yeah. Even poem.
Johannes Castner:It, it, it's true that it's happening, but I, I don't see, I still don't see the value of it myself personally. I feel that music is something that we do. It's, it's like almost, I feel that if you see yoga robots, you start saying, okay, it exists. You can make it, you can build robots. They do yoga. But I don't know what it means. You know, the meaning is a bit, is a bit, uh, elusive to me.
Umesh Upadhyaya:Right. Yeah.
Johannes Castner:So, yeah, no, this is great. I, I really enjoyed this conversation and I think that, you know, there's a lot that you are doing in Nepal that I, I think, you know, it's, it's, it's interesting that you're doing high performance computing in Nepal. Mm-hmm. I mean, this is, uh, you know, I, I wouldn't have thought that there, um, that the need is there. How many people are in Nepal? Uh, could you, uh, could you give me that? Because that would be an interesting fact for, for people who don't know this, to relate to, to, you know, the whole story.
Umesh Upadhyaya:Let, let me check that. Out so exact is 30.03 million 2021. Census says that. Yeah, it's very, uh, it's, it's like, um, there are city areas where more people are there and then especially in the mountainous areas, there is very less. So 70, 75% of our land is mostly hilly areas, uh, big hills. And then we have lots of, um, forests. And then, yeah, I mean like the city areas, uh, are less the agricultural land and city areas are less. So it's sort of like, uh, um, there's a cult. The culture is very diverse. The people are segregated, but mostly, uh, the density on the city areas, uh, I would say mm-hmm. The hill areas
Johannes Castner:and mountain area. So diverse, like, uh, there are lot of different cultures.
Umesh Upadhyaya:Yes, there are
Johannes Castner:So, so what is the mix? What does it feel like? So when you were, let's say you are in Katmandu. Yeah. So there, there is probably the biggest city, I would assume.
Umesh Upadhyaya:Yeah. Yeah. I mean, like, it's a valley. There's a Kathmandu Bhaktapur, and Lalitpur Yes. Uh, it, it's a, it's a, it's a quite small city actually. Mm-hmm.
Johannes Castner:Yeah, so this diversity in the young is, is it a relatively young country? I think it's also quite a young country, right? Compared to, um, for example, China is aging very fast, right? We, so we noticed there's this demographic eclipse or cliff that we are speaking of when we're talking about China because of the one child, uh, policy for many years, right? And there is this, you know, aging population, basically. India is a bit less like that. So now I'm wondering to what degree are you, you know, is what, what is the age structure of Nepal? Because this, I think, will make a big difference with respect to high performance computing who gets online, right? You know, how many people will be there in the future. Mm-hmm. Um, you know, is it, is it a growing population or is it already a shrinking population?
Umesh Upadhyaya:Yes, it's a growing population, especially if you look into youths. Um, the youth carry the major population of the country, um, and then a lot of youth actually travel abroad for their studies. Um, if you look into the IT sector, um, the IT sector is growing, uh, very big, uh, here and with the current change in policies where, you know, you work from abroad, there, there are certain, uh, policies, changes that has come, um, about for, uh, bringing in. Bigger markets to Nepal. So the government has made such policies. And then, um, there is, uh, big companies as well coming to Nepal and working here, uh, outsourcing their work jobs here. Um, lot of people are working on AI and ML and, uh, mentioned not, uh, other, other, other, uh, areas of technology as well. Um, there are training institutes. Uh, there are a lot of graduates coming out of, uh, technology institutes and a lot of people that have gone abroad, uh, are serving, uh, here and the international community, like for example, HPC, Nepal and other, uh, research institutes are also driving that force, um, ahead. So, yeah. Um, it's a growing population. Uh, the youth are, uh, our energy. Yeah, we should say. Yeah.
Johannes Castner:So you are sitting between one country that's kind of shrinking and one country that I think is overall maybe stabilizing, I would say India. It seems to me it's starting to stabilize. Um, and so you're still a growing, uh, population. How is that with economic growth? What is your economic growth, uh, looking like?
Umesh Upadhyaya:Yeah, the economic, uh, growth is, uh, better. Of course the covid situations, uh, led to, you know, some problems, but, uh, we're reviving again on that. Uh, so the government has come with a lot of, uh, policy reforms, uh, on how to increase the, uh, the economics, uh, economy growth and bringing in, you know, a lot of, uh, lot of, like I said, lot of, uh, it sector jobs to, to markets, to Nepal. Um, And then increasing the industry here itself. Um, on the technology parts, uh, there are industry actually, manufacturing industry also is growing here. Some, some manufacturing, some car and some bike manufacturing companies. Um, but it, it's mostly on the technical and also the, one of the biggest is the, one of the biggest is the, um, is the tourism. And now it is going to be, in very few years, it's going to be the energy sector, the green energy, the hydropower energy because, uh, and the countries plans to sell it to India through the neighboring countries, through India, China, and Bangladesh. Yeah, in few years there is going to be, uh, um, very much or abundant with hydro electricity. What Nepal will be consuming persists the level of products and cheese. It's going to be very large. So there will be a lot of, uh, green energy for, um, you know, and I, it'll be a great, uh, great, uh, uh, For the industry, like, uh, you know, we are talking about blockchain industries, we are talking about super competitors. We are talking about quantum computing. These are going to take enormous amount of energies. So yeah, I mean, with energy sector, if you look into Nepal, could be a platform or place where, you know, people can, um, start off with, uh, bringing, uh, at least, you know, look into that aspects or prospects and or a possibility if, uh, this could be a, a place of area to work on bringing technology. And also when we are looking to the green computing part, I think that's easier even for the, what's happening globally in terms of climate or weather.
Johannes Castner:So you could do a lot more computing, let's say you could move some computing from some more, from a more polluted area or from a, from an area where electricity is more polluting to Nepal. Yes. So that the computations that are being done are less polluting. Right. So that is actually a really interesting prospect for, for Neal. Yes. Yeah. Yes. That's okay. So you're becoming really kind of a high tech society then. So you'll be, you're moving from a, from a mostly tourist driven society to, to then a more high tech society, but at the same time, probably you want to preserve also the tourism, because that's, obviously you have a beautiful country, everybody wants to see it. You don't want to trash it. Right. So yeah.
Umesh Upadhyaya:That's, that's, that's very much true.
Johannes Castner:Yeah. So it's a, it's a, it's a, it's a delicate balance, right. Because in, in California, for example, we lost, uh, 95% of wildlife, you know, it's, um, a lot of the wildlife was lost, and it is really 95% or 90% of wildlife. It still has a lot of wildlife. So it's, it's incredible when, when you think about what there was, you know what, there must have been before it all started. So, uh, I think you would be very well off. Not to do, not to repeat that, you know, in, in, in your country, because Nepal is one of the, you know, one of the most beautiful places on earth. I, I know about that. I had friends who went there and so safeguard that. And maybe you can use AI and even high performance computing for the tourist, uh, industry as well. Do you, do you, do you think that there are some applications in that area?
Umesh Upadhyaya:Um, I actually see a lot of potential in that area, especially, uh, when we look into, um, When we look into the Tourism 4.0 concept, uh, the Tourism 4.0 concept is a concept that brings in all kind of, uh, blockchain, artificial intelligence, machine learning, um, and then other tools, uh, on how to, uh, better or for the betterment of the tourism industry or the tourism market. I think that would mean like, uh, we had a massive earthquake, uh, uh, a few years back, and then that led to a lot of, you know, disaster to the, uh, to the, uh, to a lot of, uh, old heritage sites and buildings. And preserving them has been one of the challenges, but still, uh, they're being rebuilt. So I think having, uh, these kind of places, uh, these heritages, um, You know, taking the 3D images of them, uh, also evaluating on what time they were built or what kind of bricks or what, what kind of labors were being used, sort of those datas and the availability of datas, those kind of data was, would mean like we could prevent, uh, from, you know, building a concrete wall and then going further down. We could also, uh, you know, bring those heritages back to life as they looked, uh, previously how it, they were. So, one qu, one area could be using those 3D models, presuming those tourism, uh, heritage, uh, heritage sites. The other could be like, the data itself is enough, I guess, uh, for example, So on the mountainous, I'm on the mountainous region. If you look into, uh, that, uh, people come here for checking and then all sort of things, and then making that into a big, uh, market size economy. And then using the virtual realities, uh, and then sort of things to actually for people to take that adventure when they do not. So many people might not be able to travel to mountains, but they want to experience that with a, with a virtual reality probably, right? And also experience Nepal, uh, the, the, uh, the scenes around the valley and then maybe go somewhere virtually. So I think that landscape all sort of, you know, data, imagery data would really help, uh, in bring that, uh, or uplift the tourism industry. I think that, that, that would be an area where artificial intelligence or use of high performance computing would be really to look looking further. Fantastic.
Johannes Castner:So, so a lot of different applications also probably, you know, in, in combating if something, for example, God forbid, something like Covid 19 were to happen again to, to trace people who are affected from and so on. That, that is also a big application, right? For, for, uh, supercomputing and, and artificial intelligence actually,
Umesh Upadhyaya:in fact. Yeah, I think, uh, that, that was done by a lot of industries all over the world, um, in Kathmandu university supercomputer. Also, they had done some research regarding the COVID one, so Amazon also had supported, uh, for this, um, for us, uh, on this part. Uh, actually they provided some, um, cloud computing platform for us. Uh, and then the university did actually some research on it. So, yeah. Uh, things like that, you know, the, the pandemic itself, uh, and then, uh, a lot of areas of the climate change and, uh, simulations for earthquake. And then, uh, like you mentioned, California maybe, uh, the forest fire itself, right? Um, uh, also, uh, playing that role where we could use, uh, you know, supercomputers on how we could reduce the, uh, forest fires. Uh, and then, uh, overall, you know, uh, using this data, uh, to uplift, uh, the, uh, human wellbeing, um, inspiring people further. I think that's the most important thing that we can bring further.
Johannes Castner:So let me ask you one more question with regards to technologies. You know, do you see that there's an interaction between, say, the blockchain, because you mentioned the blockchain earlier and supercomputing and maybe, um, you know, the metaverse and supercomputing. Do you see that there are some connections there? I mean, you were speaking about tourism and landscape recreation as opposed to that takes a lot of computing right? To, to put someone into these places?
Umesh Upadhyaya:I think on a competing level, yes, uh, there could be, but uh, uh, I have not done much research over metaverse and such, so I cannot say what would be, uh, the scenario, but definitely yes. Um, because, um, those applications will definitely be highly memory intensive or process intensive applications. They will be used and users will need, you know, Users will al always need, uh, you know, better output. And if we are not running 18 high performance computing platforms or some, some other, um, uh, high, high system, high end systems, then it's going to be very slow and users are not going to use this anymore. And when we are talking about these things, the, these are going to generate a lot, lot of amount of data. So if you're just talking a lot of amount of data, the big data, the chunk size and all sort of things, then yeah, uh, I do, I cannot say, uh uh, To a point where this is, this would be the use case, and this is the scenario where we would be using, but I see hyper computing, as a matter of fact, to be used over, um, you know, more, more into the geographical information systems. That is, that could be one use case, uh, where GIS and those, those, those part could be used. The other are also the non, uh, the socioeconomic part of it where hyper competing will be used, uh, like the tourism area itself. And then, uh, I mean, sorry, in the socio, uh, uh, economic part, the non, uh, uh, yeah, I think that's, uh, that's the, the, what you mentioned actually is a possible, but, um, right now I cannot actually figure it out. Uh, where, where exactly. And does it, uh, fit it? Mm-hmm.
Johannes Castner:Okay. Well, you know, That makes sense. And we we're growing and learning and working toward more and more of it. Right. So it's, it's, uh, we're still, I think we're still, would you say we were still in the early days on all of these things. Right.
Umesh Upadhyaya:That's very true. That's, at least in ai, we are in very early days. What's, what's coming up in 30 years or 40 years, or 50 years back then. But I think that's, that's going to be like remarkable. yeah, So Johannes. Yeah, I do have question also. So, yeah, I mean, like, while you formed this, uh, and your experience over, you know, previous, uh, long experiences, uh, of, of various fields of, of technology fields and, uh, other, uh, other aspects of, uh, non-technical areas also, um, Where do you see the balance, uh, in the, in the future generations coming? When you are talking about like collective intelligence, um, would collective mean, uh, that's going to be, uh, using the tools or technologies, or would that collective also mean just the collectiveness that among family members, the, the people who are less using technology to a less level? Uh, how would that cater into, uh, I mean like the, the whole idea of, uh, what we are talking here is artificial official intelligence. Uh, there are also people who do not know about artificial intelligence yet will be affected by it.
Johannes Castner:I mean, yeah. Oh, that, that's true. I think that actually everyone yeah. Will be affected by it, I think fortunately or unfortunately. Yes. Um, I think we have to really focus on the ethics, so. I think that something that has no ethics, it can be clever in certain ways, but intelligent, no, because I think intelligence is, is, is actually predicated on ethics, right? So if, if you, if you know how to do things, but you don't know what to do, then what, what are you gonna do? Right? So if, if you don't know what is the right thing to do or what is the wrong thing to do, and this is actually tricky. This is not always so straightforward. And maybe there are some, there are multiple right things to do and multiple wrong things to do, and you don't really, it's hard to distinguish between them sometimes because it's a dynamic world. We're living in a dynamic world. You do one thing, you think it's good, and then it has side effects and all of these things. So we need to apply AI for that as well. We need to find out what are the side effects, what are we affecting? How are we affecting it? And then we really have to, we have to tune into the, because, um, also, ethics is a, is a very, is a very diverse thing too. It's not one ethics for everyone, right? So everybody has a bit of a difference in how they think things should be. And ideally, AI would serve people on a very personalized level. So on their ethics, you know, so they are, they are of course, illegitimate ethics. They, they're not legitimate in the sense that if your ethics is to kill someone, uh, because they're whatever, or if your ethics is to subdue women or something like this, those are not really legitimate. They're actually considered illegitimate ethics. You might have an like that, but it's not really, um, it's, it's, it's just not legitimate in, in the sense that it affects others, you know? So if, if your ethics only involve yourself and things, Affecting you, then everything is legitimate, right? So it's, um, you know, because it's up to you what you want to happen to you and how things should affect you and how things should even interact with you. But what is not up to you is how things, you know, affect other people or how other people should be affected, right? So that's, that's up to them in a way. So can we build AI that is personalized in that sense? Now, of course there is a, a, you know, it's, it's, um, as Nom Chomsky correctly said, uh, you know, he, I often refer to him in certain questions because I think he's analytically quite, uh, power. He has a quite powerful mind. And so one thing he said about language, which is that, We all really, we have languages, different languages, but there is a, a bound around them, right? So, so there are certain things that can't be a human language. And similarly, there are certain ethics that can't really be human ethics. You know, they're, they're not really within the bounds, so it's not as, as arbitrary as it seems. So people think ethics is completely arbitrary because people have different ethics. You know, there are differences, but we have to understand that these differences are actually probably, uh, less than the commonalities that we have. So that's, so there is this big commonality that we, what we generally tend to find as the right thing to do. And if it's straightforward like that, we should always do the thing that everybody thinks is right, right? So that includes ai. And if AI is distracting us, And causing us to hate other people because, um, you see if, if you're optimizing. So this is a very important thing. So when you're asking this question about ai, one, one thing to recognize I think, is that most artificial intelligence, uh, applications are built for businesses at the moment. It could be, could be different in the future. Um, but as long as it's true that it's for the businesses, you have to align your business model. That means the way that you make money with your ethics directly. So if you, if you are optimizing something like engagement, Mm-hmm. You often have a problem because, uh, nothing is more engaging than a racist coming into the room, for example. It's engaging, uh, you know, not necessarily positively engaging, but it is engaging in the sense that you will pay attention to them. Mm-hmm. And so when you maximize engagement, what happens is the algorithms will automatically push those things that are more engaging in that sense. So that will, they will, um, increase racism in the network based because people will react to it. And that's what you are causing then. And then you're drawing some people, some people will then be attracted by this racism and you are actually creating more racists. So, so this is actually. Uh, something that I think we have to be very careful that, that, this is just an example, what I just mentioned. It's not the only thing that we worry about is racism or, uh, uh, but, but what I'm saying is that what is really a general rule is that you want to align your business model, how you make money with what you think is the right thing, and not just you, but the people who are affected. So this is a second step, right? So what, what you have to, the, the ethics that you, that your system should be based on is the ethics of the people affected by the system. And then you align your business model with that. So, so in a way, you en engage, you know, you engage all the stakeholders, other people who are affected by your algorithms, and you're trying to find out what is their ethics. And you iterate on it as well. You know, did we really hit the nail? Did we now, did we represent you correctly? Did we give you the ethics you wanted? Yes or no? Right? And then, and over time, we can learn from people what they feel hurtful and what, and sometimes we might make bad mistakes and bad things happen that shouldn't kill the whole enterprise of AI or digital culture, or, you know, we shouldn't then say, oh, this whole thing is bad because something bad happened. We have to correct for it, and we have to correct for it immediately. We have to be open with it. Uh, it is a mistake to fire someone for showing that, you know, this is what happened at Google, for example, right? This, uh, there have been several cases of, of people pointing out really bad. Situations that happened with the algorithms that wasn't really anyone's, maybe it wasn't anyone's point to make these bad things happen, right? They weren't unintentional. But then when someone points them out, please don't fire them. You know, let them help you improve the problem and actually point it out. Be open with it public, say, oh, we found that these things are happening that are bad in our system, and we are now taking these steps to correct for them. That creates trust and then creates a better system and a better society. I hope I answered your question. I mean, this is very, I could write a whole book about it actually. I'm working on something, um, in the, along those lines, but, um, you know, I hope I answered it satisfactorily for this. Yeah, that's, I
Umesh Upadhyaya:know that's a, that was a wonderful insight that you, uh, you just said. I, I, I very much agree on that part.
Johannes Castner:Let, let me, uh, just ask you to, if you could, um, give the listeners and viewers at home a bit of a, uh, you know, takeaway. What, what should they walk away from, from this conversation? What should they think about what, what is the word of the day from you? Uh, and, and ins, like maybe an inspirational statement or something of that sort.
Umesh Upadhyaya:Um, for me, I think, uh, I think in, uh, in terms of, uh, use of tools, uh, where we are currently is no, no. Previous generations had, you know, access to technologies that we right now have, and we are the most fortunate people on earth that we have such technologies and we can cover large audiences, um, in pretty much in few seconds or few minutes, a few minutes. Uh, so while these tools are there, these tools are actually the tools that we will be using, uh, even right now and even in the, in, um, uh, in future. And these tools will be, Will actually be the ones that will let us know where we are heading further in terms of, uh, wellbeing of people, wellbeing of the society, um, you know, the collective intelligence you said or not. So these tools are actually very important. The tools we will be developing in future, um, the high performance competing systems, the artificial intelligence itself, these all technology are going to converse or merge at a, at a level where we'll be using lot of data, a lot of insights, um, a lot of intelligence work, and then bring forth that, that uh, uh, that sort of, uh, intelligence to the market or that sort of solution to the market, to the people that will be oriented to catering, uh, for the. Wellbeing of the society or the wellbeing of the nations. Mm. So right from what we discuss in terms of use cases, whatever it is. Um, so when we are talking so much on the artificial, artificial intelligence, the data, um, the, the collective intelligence, uh, I think it's time that we should be more focused on bringing uplift, uplifting the humanity or the wellbeing and not, uh, focusing on, uh, you know, the world, not focusing on investing much on the, on the, on the security part of it where like the weapons or the arm, yeah. I mean it's, or sort of things, but on the humanity side, right? So I think that should be the whole idea and like we should also develop or look into the tools, um, where. The way we have devised for after engineering, like the old technology that's after engineering, we should also start looking ourselves within the nature of us, within our own, uh, self, the what, what we are. And then that would also mean, uh, inner technologies or maybe inner engineering itself, um, since we have so much of engineering already. So I think those use, use cases where we talk about inner engineering and the outer engineering merging itself and bringing that fact or bringing that to people will serve to the, uh, humanity or people's wellbeing the best. I think while SPC we talked off of various areas, um, we should not forget the very wellbeing or very society that we came from, our own nature and how we can leverage that into technology and bring that intelligence forward for us and for the generations upcoming years. I think that's the whole. Takeaway I would like to provide in this, from this conversation.
Johannes Castner:Fantastic. I, I love, I love it. So then let me just, uh, one more thing, and that is if, if, if audience members, people who are listening, people are watching, want to know how to keep in touch with you and how to follow your research and your work, where, where should they look?
Umesh Upadhyaya:Surely you can always look into, uh, LinkedIn, um, uh, my LinkedIn profile, and then you can always, uh, write to me at, Umesh, U M E S H hpc nepal.org. That's my email address. Uh, you can find me on Facebook, Instagram, if you go by my name. Umesh Upadhyaya, and you might see some mountains or some pictures in, uh, and then you'll, you'll find me there.
Johannes Castner:This show is published every week on Wednesday at
2:00 AM in Los Angeles, 5:00 AM in New York City at 10:00 AM in London. Next week I will be speaking with Gisel Velarde and we will be speaking about her up and coming book, the AI era, among other topics such as data science, ai, and particularly AI and music. Please leave your thoughts in the common sections if any of these topics interest you. Also, please let us know what your thoughts are about AI and music. Should AI make music? Should AI make art? How about the metaverse or the blockchain? Please let us know about all of your thoughts that you may have about the topics that we cover in this show.
Gissel Velarde:Ada Lovelace, uh, who's the first programmer, uh, she thought that analytical machines, um, could create or could be used for any complex task including creating music. So she envisioned the capability of, uh, intelligent algorithms to create music that was interesting, that also she related, um, artificial intelligence with music.