Kiki (K): Hello all! Welcome back to Kartini Teknologi,together with Kiki and Galuh here. We have Radita Liem with us today, she is a research assistant and PhD candidate also at the university RWTH Aachen University. I’m not sure what the abbreviation stands for, so maybe we just ask directly to the Dita. Hello Dita!
Dita (D): Hello!
K: Are you now studying from home?
D: As a matter of fact, for my doctorate program, I’m already working on the research. And because, yes, all computer scientists only need laptops with the internet, whether it’s working from home or not, it makes no difference.
K: So it’s just like the usual?
D: Yes, about how to mention the name of the university, you can actually say it Indonesian style, RWTH is actually more correct and many Germans prefer the pronounciation. If you explain it the American way it gets too long. People in Germany pronounce it RWTH the Indonesian way.
Galuh (G): Oh so it’s the same way of pronounciation?
D: Yes, so pronounciation in German and Indonesian are almost identical, so for Indonesians learning German may be relatively easier than native English speakers.
K: So can you speak German now?
D: Enough to survive, to order food and ship.
K: When did you move to Germany for your study?
D: I started my study in mid-2019, hmm wait is this the introduction or…?
K: It’s okay.
D: Okay. So before that I studied in Jogja in Indonesia, at Universitas Kristen Duta Wacana (UKDW) majoring in information systems. After that I worked three years or more at a Japanese company, called Adways, on internet marketing and games. Then I spent a few years at Traveloka, then after that took my Masters in Estonia in the field of computer science. There I also worked—I was quite a workaholic—at one of the biggest research centers in Estonia called STACC. Then I also went to the Estonian startup unicorn called Taxify which later changed to Bolt. Then after that I’m in Aachen, doing my doctorate studies.
K: Wow, it’s quite a long journey. But, let’s talk about how you were interested in technology. Like the beginning where you started to think that “I think I like technology. I would like to continue studying computers.” When did you start?
D: I’m pretty pragmatic, though. I saw more job vacancies in computer science, they need a diploma and the cheapest major is information systems, so I took it. Sorry it’s not very inspiring. Laughs
K: Laughs. But that’s okay though. For people who really pursue to enter the technology industry for the money, I think there’s no shame on that.
K: Does that mean you already want to enter the technology industry from high school?
D: I see a lot of potential, though. For example seeing the point of sale at the cashier, as a small child from the beginning until now I still have many questions. So I’m curious, who made this? Why is it like this? Back then, no one could answer. And this wasa very strange world but I saw its potential, the internet, computers, and so on, what is this… the future of software that cannot be stopped, it will definitely be very important for us all. So I became very pragmatic. Oh, I have to work in a place that many people will need.
K: Most of Kartini Technology’s guests are usually interested in technology after they got introduced to the internet. And Dita herself was talking about things from everyday life, right, this point of sale system at the checkout, who made it. I didn’t think of that as a kid. So, it’s really cool that you already noticed things like that. Cool!
D: Hmm yes. Thanks. Laugh. Yeah, the thing is when I was in middle school we had Pascal’s lesson. Then I wondered who made this. Why do I have to work on this? Then, after I thought about it, I relized that oh, the cashier was made using Pascal or COBOL. Then, it turned out to be interesting too. Then, in high school there were friends who entered the Indonesian computer olympiad team, they had questions that were fun like puzzles. I thought the Olympics was really fun, doing tests with puzzle quizzes like that. Well, from there I found out that his world turns out to be very dependent on software and very dependent on computers.
K: Where is your middle school that taught you Pascal?
D: In Semarang.
G: I also learned Pascal in middle school.
G: Seriously. But quite late in the year, so after Excel and stuff we learned Pascal.
K: Middle school as well?
G: Wait, I forgot was it middle school or high school.
K: I think it’s quite hard core if they taught you Pascal in middle school.
G: I don’t remember, but the lessons were simple, like we just had to make a calculator.
K: Oh wow, thst’s so fun! Anyway, you went through bachelor, master and now doctorate studies. So does it mean you enjoy academia?
D: Hmm, not really. I had a break right between S1 and S2. I started to love the world of research and school in general is just when I got to graduate school. When I graduated from S1, the story was not inspiring at all. After that I tried to work, but I saw that I was getting more curious. At first it was only a cash register, then I was curious about more complicated systems. Oh, how come you can, when I worked at Traveloka, how come millions of people can access at one time, then what kind of connection do you need, what is the team working on it like, what kind of knowledge is needed, I was very curious.
G: Because working is a new opportunity for us to do projects or work on tasks that have very different scale than what we usually do in college. In undergraduate studies you might only work on toy problems. When you are at work, for example at Traveloka, the scale is definitely very different and it opens up a lot of new opportunities.
D: Yes I think that’s right.
K: You said you started getting interested in academia in your postgraduate study. What was your focus?
D: For my master’s, I focused on the Distributed System. So this interest in distributed systems started when I was working in Taveloka. I saw how to manage traffic from various millions of people, many systems are involved. I felt, oh, I really don’t understand what this is. Why can’t I scale like this and so on. Then I was quite tired of working, I was ill for a long time. Then, ah, I thought I have to rest. But I don’t want an empty CV. Then I started looking for master’s programs. And I looked for master’s, I was deliberately looking for what makes me happy, to satisfy my curiosity. So, I decided to just take computer science.
K: I see. So you didn’t plan to take master’s?
D: No planning at all.
K: That’s cool! But you had scholarship too, right?
D: Hmm, yeah. But no Indonesians knew about the scholarship. laughs
K: laughs. But it’s true it seems very rare for Indonesians to know Estonia.
D: The story about how I know Estonia is even funnier. So when I worked at a Japanese company, there was a client, he moved to Estonia. Then, we are still friends on Facebook, so he updates that’s he’s now in a Venture Capital in Estonia. Oh, Estonians are like this, technological innovations are like that. Because of that, I got a first-hand experience from people who have experienced it themselves, who once lived in Estonia and know the situation of how they are very advanced in the field of technology. Then, I browse around and I found this scholarship.
K: Oh my god. Geez, so actually there was no intention from the beginning too. But yeah, that means Estonia actually has a lot of potential. The technology is already quite advanced there. But not many people know about it yet.
G: We also learned about how Estonia is already technologically advancedf from one of the Kartini Teknologi guests too, in episodes 3 or 4 huh?
K: Mbak Ocha, right?
G: We discussed Estonia too, because her start up is already registered in Estonia. We were also amazed when we hear about Estonia, there’s this, that, like, wow!
D: So I know this Estonia because of the Japanese who became Venture Capital in Estonia, right. He is still active until now. And we got a chance to meet actually and it was quite funny. So, he gave me a startup event pass just like that.He said “Oh, sorry. I can’t come. Why don’t you come?” so I came, right. Then I was treated like a VIP. There was my friend who happened to be on the committee there saying, “wait, how come you could enter with this investor badge?” laughs
D: Yes, so he told me that in Estonia there is already digital signing, and then all online learning is already working. So, in the midst of this pandemic many articles wrote: “wow, this seems to be the only country that is ready to be locked down.”
K: laughs. Yeah, I also hear that the bureaucracy there is already digital-based, right?
K: Interesting! You had a lot of different experiences in different fields, like in Adways, it was more about games. Then, when you were in Traveloka, what did you work on? Web and mobile?
K: Yes, and in Estonia, you worked on the data field as well. Based on previous experiences, what have you learned?
D: Hmm, valuable experience, yeah, after all this journey… I think I’m already old enough to give advice. Many students who study IT, who are still pursuing their undergraduate dgeree, most of the complaints that I hear is “why there’s a mismatch between the industry and what is taught in school?” Well not really. What is learned on campus is actually the fundamental things that are long-term in nature. So, you want to work or school later or even ten years later, it’s all still the same fundamentals. And I proved myself from university in Indonesia until I moved to Germany, with different fields. The fundamentals are still the same, algorithm, then, database management is the same, what else? Network Engineering? Yes, those are the most important things. Whether you’re working on a laptop in an internet cafe or a supercomputer system. It’s just that the level of complexity and the level of abstraction will be different. So don’t just look at the surface but start to see the fundamentals.
K: I’m curious now. Would you say that people need to have formal education to get into the technology industry?
D: Hmm, this is rather difficult to answer. There is formal education in which there are parts which you can only obtained at the university. For example, like supercomputers like this, most of them are university infrastructure. Then there’s more, the advantages of formal education are you will learn topics that you don’t think you need, but you are forced to learn. For example if you’re like “oh, I just want to be a web programmer, really”. So you only focus on programming, right? Then, you don’t study formally at university. There are actually many people who are really good at it. But, there is a blind spot like “oh, there are computer networks as well. Oh, if it’s on a desktop it’s like this, there’s a difference. So, formal education usually covers a wide range of topics. If you want to be a specialist, you have to take other courses outside of it or practice yourself.
G: Do you have courses that in undergraduate, you couldn’t think of how you could use these courses, but in your master’s or doctorate studies, you realized that oh this course is very important?
D: Hmm. Oh there’s something interesting. So, when I graduated from bachelor’s degree there was a course called Business Intelligence. It’s learning forecasting, data mining, and so on. At that time, I took my concentration there because it was more tangible at work. Then, I didn’t think it would be popular now and the name has changed, which is data science, with the same fundamentals if I see the courses or approaches. I actually am grateful I did took the class.
G: I see… interesting!
K: Let’s start to talk about the main topic, yes, about High Performing Computer (HPC). I think it’s really exciting. I think this is the first time I heard about HPC. It seems like quantum computing has more hype, so what is HPC actually?
D: When asked what High Performing Computing is, actually it has been used in all the things that exist now including internet networks around the world. Then we often get news like “oh, there is climate change and so on”. That’s actually the work of researchers who process the data using a supercomputer system or performance computing, and most recently I’m really excited, a lot of research for the potential of medicines and materials that can be used for disaster management like this COVID-19 pandemic, they are actually run on supercomputer machines and of course the science of those is in this High Performing Computing field.
K: Oh, so is HPC equal to supercomputer?
D: Not necessarily. When you mention supercomputer people would think there are many server racks in one room.
K: Yeah that’s right. That has been my imagination all this time.
D: And bitcoins that have been popular recently, actually we can categorize them into High Performance Computing. We have a big problem, then we chop it into small pieces, and they work on each computer.
K: I see. So, the whole system of blockchain is actually supercomputer?
D: Yes, you can say it’s like that. More generally, it’s a distributed system. So to arrange these computers so they can work together and also their performance–if one computer plus one computer is still two computers, not one and a half–that’s the knowledge we learn in the field of High Performance Computing.
K: I see. But that means the base is still a classic computer, right? In quantum computing there are a lot of dimensions, right? It’s not just electrons, so if in a supercomputer, do you still use electrons?
D: Yes, still the same. For quantum computing, unfortunately only IBM and Google have the so-called quantum computing. Most of the quantum computing that you hear on the internet and publications are simulations of quantum computing that is run on supercomputers. So, ordinary computers that we know today, simulate what happens in a quantum computer.
K: Okay. So there’s a difference between quantum computer and HPC.
D: Yes. But quantum computing is a subset of High Performance Computing. Because the focus is High Performance Computing, performance. The performance of this computer can be used for computations that are more difficult, then it can be used for things like, say your Zoom is slow, you need to increase the bandwidth or something. Or the algorithm is improved so that the connection is smoother, the calculation is faster.
G: Well, speaking of performance, I’m curious anyway. What metrics are usually used to measure performance on HPC itself? There are a number of metrics that we often hear. But is there a specific metric, for example, for measuring computers whose performance is much higher?
D: So, programming that is usually studied at universities, rarely has lectures on parallel programming. So, in HPC, one of the big fields is how to change these programs so that they run in parallel. Well, there is such thing as parallel efficiency. So, like I said. You have 2 computers, run at the same time to calculate something. One computer and one computer, you want the performance to be 2 right? In the real world it’s not like that. I’s one and a half. Because there are networks, or the algorithm is apparently not effective for combining the two. So it takes a lot of understanding both in terms of hardware, then in terms of the software itself, and in general how this computer works. Including the devices in it.
K: So, for example. I am a businessman, so I want to make a supercomputer.What is needed to make one supercomputer?
D: In terms of cost or how?
K: The system architecture.
D: It really depends. What is it need for? So, in this world there is a list of the 500 fastest supercomputers in the world. Some of these supercomputers are general-purpose. So, it can be used to calculate climate change, some are for medicine research, some are for astronomical research, and some are very specific. If the business is very specific, for example, there are companies that process weather data, maybe the costs are relatively lower because the focus might be more on hardware for data-storage.
K: Oh, okay, so it really depends on the usecase right?
D: Yes, it really depends on the needs. And the needs for universities are also different. Usually it is enough to show students how it works like this, it can be used for crafts, even on several campuses here we have a program, this is like a lecture, students are given devices such as Arduino or Raspberry Pi, then told to assemble themselves. And then the performance is measured. How is the performance, if you run parallel programs on it does it work or not. Or your program needs to be adjusted to hardware.
K: So it’s like simulation?
D: Yes. So can you call it as a supercomputer? They made them out of just Raspberry Pi.
K: If so, anything that we assemble using supercomputer methods can actually be called as a supercomputer right?
D: Relatively, yes.
K: Okay. That means there are no limits, for example the supercomputer must have a minimum performance this big, for example. There are no specifications like that?
D: There shouldn’t be. Because I have read a research about how people use clusters that are made using Raspberry Pi. Then they do the measurement with a large computer cluster method, yet it also worked well.
K: From what I hear, the price of a supercomputer is really expensive right?
D: Yeah, the fastest one is 350 million USD.
K: 350 million USD. Wow!
G: Tertawa. Wow!
K: What’s your current focus for your research?
D: I’m now in performance measurement and portability. Especially for I/Os and file systems. Yes, that’s what I said about parallel efficiency, how can these programs run faster according to the hardware they have. The theory is like this, but why is the practice is different? That’s the focus of my research. And my focus is on I/O, on the file system. So, if you pay attention now most laptops use SSD, right? Yes because it is faster. Actually there are also many other tweaks that can be done to make it much faster.
K: Oh, okay..
D: Yes, in terms of the program, or how you do the programming, programming language and so forth. Then, what’s interesting is portability. Portability is easy, if I run a program on a computer now, on AMD and Intel. The performance is different. Now how to make a program that can run on a lot of hardware, but the penalty to the performance isn’t too heavy?
K: Does it mean making standardization for various hardware? How does it work?
D: There is standardization. In my research group there are several friends who are on the standardization committee. There are also those who focus more on the best practice, how to make a portable program? Including, how to measure it. If I could say, this is a portable program, what are the rules, the standard?
K: Oh .. That means looking for patterns on how to make portable software right?
K: I see, I see. So you’re mostly working on hardware aren’t you?
D: Mostly software actually. We talked about how hardware is something that we can control. We can also fix the software so that it can follow the hardware’s capability, in theory. So why doesn’t it currently have the performance according to the theory from the hardware vendor?
K: Oh, I see. Means that there are many factors in software that cause it huh?
D: Yes. So these two factors must be balanced.
G: From the software side itself, what programming language is usually used in the world of High Performing Computing, to tweak the algorithm or something?
D: Most of us here use C, or C ++. More often C.
G: I see. Okay, okay.
D: Then, most of the other support goes to Fortran, because the supercomputer users are mostly scientists. Physicists, weather experts, economists are also there for stock market forcasting, they also invest in High Performance Computing. Mostly in Fortran because this language is suitable for scientists. So, for people who work at HPC, mostly we have support for C and Fortran. And, currently Python too.
G: But, I don’t know the details, I mean, many people say that Python is more high level than other programming languages, right? Is that a problem, in the HPC field itself?
D: Oh, yes it’s a problem. Laugh. Yeah, so what I said earlier, most of these users aren’t programmers. At all. They are physicists and chemists who study programming. To make it easier, they use Python, then, so, can calculate the formation of molecules accurately. Then after that, how about the performance? You know, it’s dark. Laugh. So, most of the projects that come into my place, what is the optimization strategy of this old program from other people. So that it can run on the latest system, or it can be moved easily to another system. The problem is, most of them also have programs that are involved in one of the university clusters, then other universities want to use it, they can’t.
G: Hmm, yeah. Especially each programming language has its own pros and cons.
D: Yes. We make an analysis. Oh, this can be moved, this way. Performance can also be improved, so that, for example, there is a machine learning program that uses python, and this is very slow. No matter how many nodes you give, one node, two nodes, ten nodes they make no difference. We make an analysis, what is the strategy, is there something wrong with the way the parallelization is done, is there something wrong with the way the program was written, then after that we make improvements that can be done by the scientist. You see, I’m just a computer scientist myself, right? If asked to compute the movement of molecules, I don’t know how to do that.
K: Maybe it’s still a long way. But, do you have any plan after graduating from your doctorate studies?
D: Hmm, after graduating fromit is clear that I want to add more experiences in this field. I find this field very interesting. That is it. I’ve been following my owncuriosity so far.
K: So following your instinct so far?
D: Yes so far it’s like that.
K:But if in Indonesia alone, have you ever heard of HPC in Indonesia? Anyone who does research or focuses on it. Is there any?
D: LIPI is heading there. I met with a researcher from LIPI, they’re currently studying and we are in the same field.
K: Cool. Are there other topics that you’re studying besides HPC?
D: Hmm, I’m more detailed now. My focus was more on the file system and performance. But the hottest thing right now, and the one I have a project on is to improve performance in distributed machine learning. Everyone wants to make machine learning, but what if training the data takes a year? So right now I’m working on that field.
K: Interesting! Ok, we’re shifting a little bit. The theme of Kartini Technology itself is that we really want to encourage girls to learn technology. Well, I myself see you also care about supporting more women in tech. According to you, what causes the lack of confidence for girls to pursue a career in technology?
D: I personally think it’s more about social pressure.
K: Ah, okay. Yeah…
D: I once came to a hackathon right, alone in Indonesia. People’s comments were like “wow, that’s great. A girl comes alone to a hackathon.” It’s like, dude, what’s the problem?
K: Yeah, people still seem to be amazed to see girls in technology.
D: The history is actually the other way around, software used to be women’s work.
G: Yeah that’s what I’ve been thinking about too.
D: Yes, the creator of the COBOL language is Admiral Grace Hopper. That’s why there used to be a lot of great female programmers. How come there aren’t many? Yes, you (boys) take the job.
D: So back then, boys usually worked on hardware.
K: I see!
D: Girls only did the “softer” work, like typing, secretary’s work.
G: Yes, from what I read, people used to consider software as administrative work. The cool thing was hardware, where the guys are. Until, there came a time where people realize that software is the next big thing. Well, then the guys took jobs that were actually pioneered by many girls.
D: True. laughs
K: Ah, I see. I remember, when I saw the film, I forgot the title, about NASA having scientists computing something for the Apollo project.
D: Oh, Hidden Figures isn’t it?
K: Ah, yes. It’s about black women who work at NASA right?
D: Yes, yes.
K: Yes, it’s really great. When I saw it I was like, wow!
D: Yes, so that’s how it works from back then even until now actually. Calculations are done by scientists and these calculations used to be traditionally calculated by women. So the human computer is real. So, actually the computer is the one who does the math.
K: Yes. Especially in the past, they had to write on paper? The technology isn’t as sophisticated as right now, so it’s like, wow!
D: Yes. Actually, NASA projects are the pioneers of HPC.
D: So at one of the conferences that I had come to, the keynote speaker was the scientist who made the Mars project.
D: So, if for example you see the Hidden Figures movie where people count everything, it still exists today. But now they’re helped by supercomputers.
K: Ah, I see.
D: So, instead of counting on paper, they type the program so they can determine that “oh, this launching should be like this, the surface of Mars is like this”.
K: So it’s out of necessity too.
K: I see. Do you have any female role model in tech?
D: I think the coolest female figure in tech right now is Diane Greene. So, everyone who deals with servers knows VMware right?
D: She is one of the pioneers. So, the one who made the VMware company the first time was a girl, her name is Diane Greene. She is now on the board of Alphabet. And she was in charge of Google Cloud. Yes, she is still active. In my opinion she changed modern cloud computing, modern cloud computing is like this thanks to her work. And she is still alive.
G: Yes, she’s still active in many places as well, in Stripe, in Wix too. Wow!
K: Well, these people are rarely exposed. I guess we have too few women role models? I means if I look at it, we always go back to people like Sheryl Sandberg, someone who is already known to a lot of people. Even though there are many women whose contributions are also not less great in the field of technology. OK. Finally. Are there suggestions for Kartini Teknologi listeners who want to start a career in technology?
D: Hmm, just do it. Do not hesitate. Like, for example, I was someone who didn’t have much motivation, and I get to be where I am now because I try to keep going. Keep moving forward and keep learning to not care about “ah, this is girl’s work, this is boys’ work. If you like what you do, just do it. And try to be the best in that field.
K: Yeah. It seems that every Kartini Teknologi guest who is asked the question has the same answer: just do it!
D: True, just do it!
K: But it’s true too. It seems like girls may overthink it, maybe. So, just do it, girls! Okay, thank you, Dita for sharing a lot with us. This is really exciting. I’m sure my friends also learned a lot from this session. So yeah. Thank you so much for coming.
D: Okay, thank you!