#161 Harut Martirosyan: What Secrets Lie Behind Armenia's AI Surge?

0:00It's because that we had some heritage from Soviet times, many of the very advanced researches and very advanced engineering components for satellites were built in Armenia and we had a good institute for mathematics. We actually had two. But they say Armenia was a hub of some informatics, mathematics. Maybe that's why it is still considered, let's say, Silicon Mountains, maybe. Having the people from NVIDIA or Yandex or Adobe come here, that enriches the community, right? Of course, for sure, 100%. Hi, before we begin, I want to give a shout out to our sponsor this week, Babbel, the science-backed language learning app.

0:48I'm terrible at reading these promotions, but Babbel is something I feel strongly about because language is the key to opening the world and broadening horizons. I know because as a journalist, I've reported out of more than 40 countries around the world. and it's amazing how knowing just a few words of the local language will open doors and build bonds. Be a better you in 2024 with Babbel. Don't pay hundreds of dollars for private tutors or waste hours on apps that don't really help you speak the language. Babbel's quick 10-minute lessons are designed by over 150 language experts to help you start speaking a new language in as little as three weeks.

1:31Babbel's designed by real people for real conversations. Babbel's tips and tools are approachable, accessible, rooted in real-life situations, and delivered with conversation-based teaching, so you're ready to practice what you've learned in the real world. It's so easy to learn how to order food, ask for directions, speak to merchants, without having to consult language apps while on vacation. Plus, Babbel's speech recognition technology helps you to improve your pronunciation and accent, something that I need a little work on. Studies from Yale, Michigan State University, and others continue to prove Babbel is better.

2:09One study found that using Babbel for 15 hours is equivalent to a full semester at college. Babbel has over 10 million subscriptions sold, plus all of Babbel's 14 language courses are backed by their 20-day money-back guarantee. Here's a special limited time deal for our listeners. Right now, get 55 % off your Babel subscription at babel.com slash IonAI. That's E-Y-E-O-N-A-I, all run together, babel.com slash IonAI. Rules and restrictions may apply. Hi, I'm Craig Smith, and this is IonAI. In this episode, I talk to Harut Martorosyan, CTO of the Armenian AI company Intent AI about the country's emerging tech ecosystem.

3:03Marta Rosyan provides background on his own path to AI and shares observations on Armenia's Soviet era computing legacy. He discusses Intent AI's privacy-focused ad targeting platform, the challenges of finding experienced AI talent locally, and Armenia's orientation toward Western markets despite regional geopolitics. Marta Rossian reflects on the recent hype drawing companies like NVIDIA and Adobe to Armenia, though notes more fundamental growth is needed. The conversation offers an insightful look into Armenia's ambitions in AI. So my name is Harut, I'm 36, so I went to school in 1994. I'm from Yerevan.

3:56I graduated the State Engineering University. Then I've got my master's degree in the National Academy of Sciences. Both degrees were obtained in the direction, in majoring in mathematics, computer science, informatics, or something in between. So currently I'm the CTO of our company. It's called Intent AI. We're doing AI and machine learning, cooperating with mobile operators. We're trying hard to create the best and the most privacy respectful ad targeting technology because the ad tech world is full of fraud and black market data, stealing data, selling data, whatever, tracking everyone. So we try to build a system that's free of all that negative stuff and that focuses on two things, best possible effective targeting that you don't waste your money and privacy respectfulness because working with telcos, we're obliged to be privacy respectful and we decided to use it as our main motto as well.

5:14That's what we do. And the education, so that was before independence, is that right? Independence from Soviet Union? Yeah. It was in 1991. I went to school in 1994, actually three years after. And in between we had a war going on that I was lucky enough not to witness because I was a child and we were kind of okay during that time. Our family were not like starving. Yeah, but I remember those years as happy years actually. I was a kid. Yeah, yeah. And then the university, when you were going to the university, were there courses in AI and machine learning and computer science? It was mostly like hardcore mathematics, no AI.

6:09There were not even good courses for software engineering because it was just starting up. But now I'm sure there are good courses in the university for software engineering. Not sure for AI. Although I know that there are some specializations now, especially in the American University of Armenia. they have a bachelor's degree program for AI and data science. I mean data science. And I know that there is also a problem in Armenia. There are no like supervisors for PhD programs in Armenia that specialize on AI data science. But at the moment when I was studying there was nothing related to AI. AI was not a thing actually.

6:56AI was only researching the universities and it was like that for tens of years. My father also did PhD in machine learning and AI in the 70s. But at that time it was only something semi-fictional that maybe one day the resources would allow to do something with it. yeah yeah and back in the days when i was studying it was there were some examples of ai but it was not widespread it was not everywhere for sure so there were no courses in armenia yeah and online education was also not available for us too that much because we had problems with the internet growing up in the early 2000s i actually did a lot of caused a lot of problems to my parents just using the Armenian dial-up too much and the bill was through the roof and I was only using that traffic to socialize in the forums there were forums back in the days and also to download some materials that I would need for my software engineering, some manuals, whatever because you cannot just Google something because you're disconnected, you need first to connect to the internet So it was not something widespread, not something very available, we can sum up.

8:25And then where did you get your machine learning background? Well, I'm not typically doing machine learning in the office. I'm more like a technology person in the company that's responsible for all the technology. but I had first of all as I think all current machine learning people I just watched a lot of videos primarily the Coursera famous course by Andrew Yang which everybody I think should take because it's like fundamental although slightly outdated but still a fundamental course to take more specialized sort of courses that's all. I don't have any formal education that's related to machine learning specifically.

9:19Yeah, so the product though that you're working on involves machine learning. Yes, it's heavily dependent on AI. So what I can tell in a couple of words what it does, we deploy our product inside mobile operators premises and we require mobile operators to provide us with anonymized data for regarding the usage of their subscribers like the traffic etc and without taking it out our software processes that traffic and infers an anonymized profiles for each subscriber like this one is a male 35 years old interested in cars using whatsapp on a daily basis stuff like that and then we use that attributes to target advertisement to relevant people and all that profile inference parts is ai the second part is when you only targeting the advertisement to right people is not very efficient in advertising already so you need to use ai also to from the targeting you need to bring in only people that are genuinely interested in whatever real advertising like matching for instance uh the like banner with although you're like narrowing down the targeting with that criteria uh that's still not enough because it's too broad and many people although they fall on under your like criteria there they will not be continuing maybe they will click but they will not continue like with your product so we we're going like a step further and we try to bring people that not only click but are also interested in your product.

11:02That's also AI. There are lots of like very niche places in our product that also we use AI. So it's everywhere in our product actually. We cannot do with it at all. No tech company at the moment doing like programmatic advertising can do without AI. Right. And the ads are delivered on the smartphone screen? So that's also one of our points, that we don't introduce any new sort of advertisement. We don't send any push messages or SMS. We don't do it. We partner with ad exchanges that are brokering the advertisement. So we just participate in an auction. and whenever our system detects that someone is about to see an ad, we bid for that possibility to display our ad.

11:55And if we win the bid, the person gets exposed to our advertisement and that's it. So we don't send any obtrusive like SMS push and whatever. So we try to be good guys actually in that bad company of companies. Yeah, that's fascinating.

12:14Given how nascent the machine learning community is here, have you had trouble building a team? Well, at the moment there are lots of juniors because it became very popular to become an 80s scientist. But finding a good senior specialist in Armenia is very hard. not only because they ask a lot of salary, but also in IT overall, not only in data science, the market is not fully formed. So there are no very strict ranges of salaries for these years of experience, etc. So the range for, let's say, a mid-level engineer or data scientist can vary three times. from company to company. And that's very hard.

13:10It makes it very hard to filter out the candidates, talking to them, making some fair judgments on their salary within the scope of your company. And also, I mean, I don't know. If you're looking for a junior, you will definitely find one easily. If you're looking for a decent specialist least with matching experience, it's really hard. That's the picture. Yeah. I'm interested in how Armenia is developing and where it's going. It's very much just a very strong tech ecosystem here now. A lot of big companies. Yesterday, Adobe's CEO was here. and NVIDIA is here. I believe IBM has R &D. I'm not sure about IBM, but NVIDIA for sure they're here.

14:13Yeah.

14:17Can you give a little bit of the history of why Armenia? I think it's because that we had some heritage from Soviet times. many of the very advanced researches and very advanced like engineering components for satellites were built in Armenia and we had a good institute for mathematics we actually had two like one of them was the automate machines problems research institute the other one was the Mergelian Institute so we have some pedigree let's say in mathematics and engineering maybe that's why it is considered they say I was I cannot like I'm not a witness of that times but they say Armenia was a hub of some informatics mathematics maybe that's why it is still considered let's say Silicon Valley with mountains or what they say?

15:25Silicon mountains. Silicon mountains maybe. So maybe that's why. For myself, my father was also a mathematician. I can say, I don't know. Country wise, I don't know why it's considered the hub to be honest. Because I was not present during that times to say what happened during Soviet times etc. But I don't believe that, to be honest, I don't believe that we have something very special sort of talent in mathematics, etc. I don't believe in that kind of stuff at all. I don't believe any country has that kind of talent. It's just the market that's developed, whether you have it or not, that's the question.

16:09and at the moment i don't think that our market is very developed and we're like way ahead in terms of it if we judge the other way there is a site called stack overflow they publish statistics yearly it's like a developers engineers hub where they ask question exchange with experiences they report statistics of visits from all the countries and they allow you to query so if you query our country versus the neighboring ones it shows that we are ahead of them that's the only like number-based like indicator that i know of that we are ahead of like neighboring countries let's say yeah yeah and it's not an official statistics or yeah sure But I think it's true.

17:03The side don't lie about the statistics. And the feeling I'm getting here is there's a lot of energy, there's a lot of excitement. That's for sure. And is that because this tech ecosystem has finally reached a critical mass, There's now venture capital here, and there is the human capital. I mean, people have been focusing on machine learning. I think the excitement is more from PR, because many, unlike before, some of the Armenian companies managed to unlock some big investments, and that brought up a lot of hype. and a lot of people got excited about that and they wanted to get in to participate maybe to succeed themselves as well I think that's because of it basically it's not because of some accumulated potential that we've built over the years etc.

18:15I am actually a little bit maybe pessimistic or pragmatic but I mean that's my opinion. When I was starting I think my first job was when I was 16, I think. I was lucky enough to work with my profession or something else. And at that time, there was no indication that software engineering and IT is a very lucrative profession. Let's say that you earn a lot of money. Me and my friends that we restarted, we just liked it. That's it. And we went to our first place without asking for any money, just like, hey, can we come in and help? Yeah, that's interesting. And with now the universities offering courses, do you see the machine learning developing in industry here?

19:23For sure. Yeah. It's not only universities. There are also like private educational institutions that offer courses. Some of them are famous. Some of them are not. But nowadays, I'm 100 % sure that if you want to learn software engineering and AI or anything related to that surrounding like professions or specializations, you don't even need money. There will be a lot of places that will offer you to learn for free. Then they will take some part of your salary for one year. So there is no question that if you want to learn, you will learn. not even telling about like learning it yourself basically which which is the best way i guess i mean now you have access to internet that also boosted a lot of possibilities yeah so nowadays i don't see any obstacle that will prevent you from learning ai machine learning or software engineering if you want so yeah uh the armenia politically there are just to add sorry of course there are some limitations especially with machine learning like graphic cards they are very expensive etc but you don't start from you never start from them from from that that's a very advanced topic comparing with the people that just start learning machine learning and data science.

20:55So to start you don't need anything. You just need a computer. I think almost any will do. That's it.

21:09Politically the country is in an interesting space because you've got good relations with Russia, good relations with Iran, good relations with Georgia, but not good relations with Azerbaijan or Turkey. And yet the country is very U.S. focused. So at least in the private sector. For sure. Do you feel that the U.S. is the future or Russia and Iran? I mean, those are large economies. For me personally, I wouldn't like them to be our future, neither Russia nor Iran. They do not represent what I personally think about how my country should develop, for sure. and at the same time I don't think that going after someone is the best thing you can do the interesting situation we referred to is because it's not because our partnerships or failed diplomacy not so much because of that but more because of our own mistakes that we were making during the last 30 years having corrupt government not paying too much attention on being stronger being more useful country for others that they can trade they can partner with it's more because of us not because of our partners and technologically speaking of course US is dominating the world by their technology and like everything else and i would i think that uh culture wise as well i mean we should we need to move like build companies that have more like u.s sort of culture or european sort of culture rather than what russian companies have although i think there are some russians russian companies also have that like big ones uh but i don't think we need to go after russia by any means mostly i think although it's very dangerous for us it's a very crucial step if we manage to get away with it with minimal damage what would you mean the step meaning yeah getting closer to the u.s getting closer not just getting getting closer to others let's say getting closer to europe getting closer to us uh i think uh if we can do so without big damage we should but honestly i don't know how to do that without big damage we don't have uh we sold all our infrastructure almost all exclusively to russia railroads gas infrastructure i think power grid with the nuclear power plant i think I may be wrong with that, but I don't know.

24:23So it would be very hard to do that. But I mean, psychologically, I mean, culturally, personally, I am more on the European and on the US side of things because it's easier for me thinking that way. And for I think for 90 percent of people from my generation. Yeah. The arrival of these big tech companies here, is that changing the community or the ecosystem? I know it's driven prices up. Yes, a lot. Yeah. Well, actually, the companies you mentioned, NVIDIA, they arrived here by accident, kind of, because they acquired the company that had an office here. It was called Workfront. Before that, it was called AdTask, as far as I remember, AdTask.

25:26And they acquired that company. So now they are present in Armenia, like, automatically. This is NVIDIA or Adobe? Adobe. Nvidia I don't know if they intentionally like of we opened office here but to my knowledge they just relocated their Russian personnel to Armenia but one of the key people from Nvidia his name is Rev I think you know about you may know about him he's Armenian maybe he somehow managed to open a branch in Armenia with I don't know the story of Nvidia to be sure but I know that a lot of Russian people moved into Armenia to work from Nvidia from Russia from really big companies we have synopsis here they were when I was working in the same area they were like thousand close to thousand people

26:26I think that heritage thing maybe drove them to Armenia that they knew that these people can have potential of doing that. They were doing that in the Soviet times because they came here before all this AI hype happened in early 2000s or even late 90s. I think early 2000s. Many of the big companies have Armenian ties like Service Titan. The founders are Armenians. They are also our good advisors, one of them.

27:04Yandex is here. Yandex is here. They moved in just because of the situation in Russia, Ukraine. They just moved some of their, for instance, the food delivery thing. They operate, they develop it from here, I think. They moved from here. We know that other departments also moved to other countries because we're in the ad tech space. We work with them, we partner with them. Some of them are in Kazakhstan. some of them are in belgrade that's from the biggest ones i think those are the biggest ones we also have some representatives of very big companies but they are not like engineering offices they are not big here yeah but having the the people from in video or Yandex or Adobe come here that enriches the community of course right for sure yes also they participate in the all the gatherings meetings IT events they do knowledge sharing.

28:13I know that specifically there were some professions in the telecom industry because we work with them. There were no specialists in that area in Armenia and all of a sudden within a day there are like three candidates that arrived from Russia. So lots of gaps are now field with qualified people. It's not just like out of thin air making decisions anymore. Certainly the richer the community, the better the community, the better for the industry. They are just two related things. Yeah. And the market, however, is really Europe and the US that you're going after. Isn't that right? Our company? Well, the startup ecosystem in general.

29:09That's the thing that Russian people, when they arrived, that's the thing that they mentioned first. They say that in Russia, when somebody is making a startup, usually it's first intended to be monetized in Russia. Then they're having all that problem scaling out, changing many things. In Armenia, from the beginning, you are oriented towards West. or towards your markets that are not Armenian because Armenian market is small. Our company is also not an exception because if we put serious efforts into Armenian operations like selling our products here seriously, we'll never bring the money back because the market is way too small.

29:55That's why we focus on our countries where we have integrations with at the moment. It's Middle East, Africa, Far East Asia, and then also Europe and the US. In Europe, we also have in Poland, we're starting now with T-Mobile. And to develop those markets, do you rely heavily on the diaspora? No, not at all, actually. I mean, if we have some network there, it's good, but it's not something we rely on. For instance, in Africa, there is particularly no Armenian diaspora, I think. And we're tapping into African-Nigerian market. In the Middle East, there is Armenian diaspora. However, we don't have anyone Armenian in that part.

30:44With our advisors, we have diasporan Armenians in our advisory board, let's say. among our investors also of course there are some armenians from diaspora but it's not that but only from from investment perspective i would say we don't rely on any like technologically we don't rely on the diaspora that they will do something for us there is no such program basically if you have someone in the diaspora that knows someone the person will help it doesn't matter if the person is armenian or not So the diaspora factor, I think it's not something anyone relies on. Right. Nigeria is your first market with this?

31:36First big market, yes. Do you picked it because of the population? We didn't pick it because of the population. Our partner, mobile operator, they had like 12 or 14 countries. countries they picked it because of the population because we had no other choice so they told like we will give you a big country so that we test your product whether it can work under that big load or not yeah it did actually so we installed our product there with very it was very challenging because of the security restrictions to our surprise actually Nigerian telcos were way more strict than the others we've worked with we managed to do some changes in our technology to make it work under that constraints in that environment but after all it worked yes it's not our choice at all if it was our choice we would definitely choose a country with a more developed country because advertisement in more developed countries is more expensive and you learn even with a smaller margin, you earn more.

32:53Yeah, yeah. Are you looking at the U.S. market? Sure. I cannot disclose it yet, but we hope we have a good meeting in November. Let's see. Yeah, we're looking eventually at the U.S. market. In the U.S., if you go to the U.S., I think it's like one-shot thing. So we need to be really, really, really well prepared for that and have something to show, to demonstrate from the previous experience. Because if you fail it, you fail it for good. Yeah, yeah. The thing that kind of amazes me is that this physically small, geographically small country, not a huge population, is competing in the global market in which players are very well-funded.

34:06In some areas, it is competing, yes. Yeah. But it's not like a general competitor to the US IT. In some areas, there are companies in Armenia that can do better than other companies in the US. And that areas are usually niche areas. They are not like we do, we are always better at, let's say, AI or building, generally building some systems. But there are companies that specialize on something and they are better than the competitors in the US. It's true. Yeah. And how do you identify those niches? Which niches are they basically? Where you feel you can be competitive. And I don't mean your company.

34:58I don't know if the niches can be clustered into some categories. It basically depends on individuals that manage to build good companies that are competitive in the area. We don't have, as I say, I don't think we have any special talent in some area, let's say image processing or I don't know, parallel computing, whatever. It's just that someone managed to actually build a good thing regardless of the area. but ai like tools using ai uh we have good examples on that that we outperform the like larger competitors so if we take ai as a niche maybe there yeah yeah uh most of the companies that beat the competition in the us i think are ai focused to my knowledge.

36:03But nowadays, all the companies are AI-focused. I mean, for instance, if we consider a service-tight Armenian company, they have a huge chunk of the market, but I don't know how much they are reliant on AI. The things seem to settle down with Azerbaijan, but do you think Do you think that will limit Armenia's tech development or do you think that you've learned or survived in a hostile neighborhood? No, that limits actually and I know a lot of cases where either investors are afraid to invest or companies are afraid to bring in the branches in Armenia. that severely i think even with that like restrictions uh the community grows but that restrictions restriction exists because they're afraid to bring in their uh their companies to armenia because of that and i don't think that we are now settled really hopefully we'll settle one day of course but yeah that exists for sure so investors are afraid to invest I know cases that they asked the founders and like the core team to relocate so that they invest yeah so it's relocate to California or your California Europe Europe, US or some safer place maybe of their choice.

37:39Yeah, yeah. So that's a problem. Yeah. But it's not like a very, very big problem. I just know like three, four cases like that. Is there anything I haven't asked that...

38:01Well, maybe about like what problems I personally can see in the market. First of all, the absence of market, as I mentioned, that there is no market that is brought to some specific shape. It's always changing, fluctuating. That's a big problem. The other problem is because of that many people think that they are like the top of the world, but it's not really, they just don't have the relevant experience to say that. And companies that are starting here, there's not really the option to exit through the stock market in Armenia. I mean, the market here is still... There are legislation problems in Armenia, actually, of making the market very available to everyone because the legislation is not developed and it is very hard to protect the rights of the shareholders use with the existing laws but if an Armenian company wants to IPO the right way now is to open an US headquarters and yeah IPO there and or bring investments there that's what most of the Armenian companies do.

39:32I think like 99 % of the Armenian companies they register Delaware entity because it's easy and that becomes their headquarters in the US. Yeah. And again with Adobe buying this company here do you see that as a developing trend? to my knowledge actually workfront was not an armenian company it was acquired as a u.s company it was it was a branch of a u.s company in armenia so it's not that adobe bought an armenian company but there were some local acquisitions i think pixar taught some company I don't know the exact amount, but I don't remember that any giant company would buy an Armenian startup.