How To End The War Between Translators and MT – Oksana Tkach From Metamova

•      Was this content helpful?

Was this content helpful?

MT is getting better each day… what is the future for translators? Oksana Tkach, the CEO of Metamova, is working on a solution where translators and AI work in harmony. Inspired by blockchain, she shares her idea of a distributed network with MT that belongs to no one.

Oksana is also an expert on NLP. Here’s your chance to learn more about natural language processing and its practical application in localization. We also discussed:

  • Why big localization companies are going to collapse
  • Creating chatbots for e-commerce
  • Why you should be careful during the AI hype
  • Rule-based vs statistical NLP
  • Why translation companies need NLP
  • MT engine in a shared economy
  • Why BlaBlaCar overpaid their translators
  • Quality assessment through consensus
  • Who the hell is… Martin Eden?!

This is episode #24 of my speaking / social interaction practice, also known as The Localization Podcast 🙂 #localization​ and #translation​ insight delivered to you by the power of voice.


Andrej Zito 

I have a new guest here this time is Oksana Tkach. Hope I’m saying this correct. from Ukraine.

Oksana Tkach 

Yes you are.

Andrej Zito

Okay.

Oksana Tkach 

Very nice.

Andrej Zito 

You’re the CEO of Metamova. Right.

Oksana Tkach 

Correct.

Andrej Zito 

So how is your day? And where exactly are you located?

Oksana Tkach 

Oh, well, I located in the Ukraine, that’s in western Ukraine. I am in my office, which is also my home. And I am currently under quarantine. Yes, I’m trying not to leave this place. So a little bit depressing. But yeah, getting a lot of work done.

Andrej Zito 

How’s the situation with the with the virus in Ukraine?

Oksana Tkach 

It’s actually really slow. Because I’m, it just less people travel here, I guess. So we had like, one death and three confirmed diagnoses. So super small, but we are still taking precaution because it’s seems like the right thing to do. So yeah, we are just like all canceling all events. And it’s okay. Everything’s fine.

Andrej Zito 

Yeah, you told me about your living conditions, when we had our previous call together before this interview. So how did how did that happen? Like how did you end up having your office and bedroom in the same place?

Oksana Tkach 

It’s such a weird situation. So I was looking for an office. And I couldn’t find the one that I liked, like for my team, right, because I have a team of people. But I found this weird place the spirit space that was basically the owners, they bought all of the apartments on the one floor, and they merged them into this giant office. But it just needed one room, but they were looking for to rent a lot of this space to someone. And then weirdly, my roommate was moving to Berlin. And so we were losing our old apartment, I was like, I’m just gonna move in here because it looks really cozy. And so basically, I have this studio apartment, and one corner of this floor. And then we have a separate entrance for my guys, and I will to office. But then they are also connected by another entrance that goes right into my bedroom. So literally live. Like I can hear them come into the office in the morning while I’m still in my bed. It is so bizarre, but also kind of it’s also kind of really handy, because I just don’t stop working. And I’ve I’ve just done so much work in the past two months.

Andrej Zito 

Yeah, I know. I know. You told me that usually, like, wrap it up, like around midnight. Time, right?

Oksana Tkach 

Yeah, it’s not super healthy. But that’s how it ends up being. Yep.

Andrej Zito 

Right. Do you remember how did you have the limited space when you were growing up?

Oksana Tkach 

What do you mean?

Andrej Zito 

Like when you were a kid like with your family? Big house small house.

Oksana Tkach 

So we we lived You know, every time you see a movie about Eastern Europe, you see those like block buildings. That’s that’s where I grew up. Like my parents got this apartment for free in the 90s. But it was like two rooms and a kitchen. So I had to share a room with my sister. So it was pretty crowded. And this is probably the first time in my life that I have this huge space all to myself. And it feels really good.

Andrej Zito 

So do you remember like what kind of kid you were in high school?

Oksana Tkach 

A kid?

Andrej Zito 

Yeah. Were you like a popular one? Were you the nerdy type?

Oksana Tkach 

No, these questions are really triggering me.

Andrej Zito 

Why? Explain.

Oksana Tkach 

I don’t know. Yeah, I wasn’t popular. I was just very kind of I was a nerd. I like I like YouTube. And I

Andrej Zito 

guess you’re young because when I was growing up, there was no YouTube. The internet was barely starting.

Oksana Tkach 

So it was like, YouTube was starting when I was like 13. And I got good internet at that time. And I was like, Okay, I have discovered internet. That’s where I’m living from now on. Because it just felt like the just felt closer to the way I was feeling than the culture in Ukraine at that time, or the people in my school? Um, yeah, and just kind of lost. And just in need of some direction.

Andrej Zito 

Did you think that the you were like the forward thinking person in Ukraine? Is that what you meant? Like?

Oksana Tkach 

I was just dissatisfied? I don’t know if you would call it forward thinking. It’s just it kind of drives you to look for something else. I just, I felt like something was wrong. Am I forward thinking right now? I don’t know. I like to solve problems. I don’t like when things aren’t working. Well. Who knows?

Andrej Zito 

So what were your interests? Like in high school? How did that? Did you already have like some interest in like, language or linguistics?

Oksana Tkach 

Well, I, so I discovered YouTube. And I was like, I want to understand these people. So I just learned English. I think I was fluent in English, at like, 16, just because I was watching all this stuff. And be like, I will not stop watching this until I understand what they mean. So that was cool. I didn’t think I’m like, super linguistically gifted, I think I was just really wanted to learn English, because it was somehow important to me. And then, but then, I think it was around when I was 16. I found this article on Wikipedia about computational linguistics. And which is, which was basically like, you can quantify language, like, you can take language and explain it in numbers. And I was like, Oh, that’s so cool. That’s what I want to do. And so I decided that that’s what I was going to do. And it took me about 10 years to start getting real projects. So it was, it was I was 16. Now I’m 26. And I’m getting like really cool, big projects right now. So 10 years of studying, but yeah, I think I got there.

Andrej Zito 

So yeah, I think I think we can like break it down a little bit like this. 10 years. So if your journey. So what are like your options, like when it let’s say, your high school student, and you find this interest, like in language? Like, did you have a chance to study something related to it? or What did you do like right after high school?

Oksana Tkach 

Yeah, it was a problem. Because I mean, there were programs in computational linguistics in what is right now called natural language processing. But like they they were in Stanford, or in somewhere in Europe, and they cost money, and we didn’t have any money. So the opportunity was kind of limited. At that time, I did not believe that I could even like get a visa to get out of Ukraine. There was just like, not in my mindset. So I decided, and I remember thinking, well, if I can’t, like if I don’t have the same baseline, right, as other people in the world, then I guess I’ll have to prove I have to do something to prove that I deserve the same treatment, even though other people don’t have to prove that. So I was just like, Okay, I’m just gonna work hard until I get to the same level. So I, eventually I decided to go study translation. That’s how I got into translation. Because I felt like well, I can learn programming at any time, right? But linguistics is way more kind of innocuous. And it’s like, there are art courses online for linguistics. I wanted to get like a really classic education in linguistics. And also, I thought, well, if I’m already fluent in English, then it’s going to be easier for me. And it was so that was, that was a good move. Because I didn’t have to study as much. Yeah, so I basically ended up studying linguistics. And I got a job and basically translating. So my last year of college, I started to translate for the for this agency, Google products, Google localization. And that’s how I found out about localization. That was really cool, because it’s kind of like, technical language. And I also started to learn Python, the programming language. So if you want to, if you’re interested in this field, I recommend highly to start with Python. And yeah.

Andrej Zito 

How did you how did you discover python going from translation because like I also discovered Python, because, you know, recently everybody’s talking about machine learning, deep learning and all these things. So like, the first course that I found was like Python like patterns. The The one thing is this kind of similar to how you discovered Python and all these programming, and how it relates to linguistics.

Oksana Tkach 

Yeah, so I was just kind of researching, where should I start, if I want to do computational linguistics, and I found somebody recommending Python and probably r. and Python is like, it’s, it’s really kind of readable, because it was created to kind of read as a human language. So it’s great for somebody who’s kind of into like, has a brain that’s more into art, it’s a bit more clear. And it also better just the way the language is structured is better for processing text. So Python is great for language, natural language. But then when it comes to, but then so machine translation and data science, they are very much connected to computational linguistics. So it’s kind of like almost one field. And it’s kind of easier to do machine learning in Python as well. So that was like, I don’t think all of this reasoning that like all of these reasons that I know. Now, when I went into that decision, I was just kind of like, randomly trying to do something. And I eventually did the right choice. made the right choice. Yeah. So yeah, just started doing something anything.

Andrej Zito 

How did you like your first translation jobs? Did you feel like that’s the thing that you want to be doing? Or was it just like a step to help?

Oksana Tkach 

Oh, man, I’m not the best translator.

Andrej Zito 

That’s okay. That’s okay.

Oksana Tkach 

Sometimes when I’m like, inspired, I can do something cool. But I am not consistently, like, it’s very rare for me that I am translating. And I am like, in the flow for hours and hours, that happens for me with programming. But when it’s because I worked as a translator for probably like, at least three to four years. And I did like big transition jobs. But it was almost like it was I was just kind of forcing myself to do it. Because I had to make money somehow. But I just feel like there are so many people that are passionate about the process of translating. And to me, it’s kind of like, it’s, I mean, I know it can do it. But it’s not something that is that excites me. And that is not to I mean, I admire people that enjoy translating.

Andrej Zito 

Yeah, I know, I know exactly what you’re talking about. I have the same thing with Project Management. Like my career started as an engineer, you know, so I always saw the project managers as the gods, you know, who could make all the decisions, and they were dealing with clients, you know, and they were organizing and planning. Like, once I started doing it for like, couple of years. And once I kind of like, feel like I figured it out. It just lost like the the excitement for me. Because to me, it’s more like a production line. Or not much exciting is happening every time you get a project. Because basically do the same things just with like different source files.

Oksana Tkach 

It’s not as mysterious anymore.

Andrej Zito 

Yes, exactly. I think I think that’s where I’m kind of like you because I also like to solve problems, because problems usually are something new. And if the problems are already recurring, which is what I have to do, I just call it customer support, because you don’t have to figure it out.

Oksana Tkach 

That’s what localization engineering is. Problems that you can’t automate. So you just have to solve them over and over again. But is it project management exciting, when there is like a big mess? And you have to fix it?

Andrej Zito 

Yes. Yes, exactly.

Oksana Tkach 

That’s the coolest. Yeah,

Andrej Zito 

that’s, that’s the, I don’t know how to call it. In English. It’s like, like, if you have too much mess, like on an ongoing basis, it’s very difficult for people to, to to keep up with it. But actually, yeah, I would definitely say that when you get a mess, like, you should look at it from the positive perspective, because that’s where you can actually like, learn something and or like, see, like where your limits are, because that’s where you have to actually start using your brain and figure things out, like how you can deal with a mess versus just doing, you know, like 1234 or five steps to get the project done.

Oksana Tkach 

Yeah. Well, what I’ve done with my life right now recently is that all of the work that I do is just fixing messes. So that is super stressful.

Andrej Zito 

We’ll get to that later. So let’s just try to briefly like cover what you were doing like after school, I have looked at your LinkedIn profile and you work for a company of companies. Most of your titles are either localization specialist, which I don’t know exactly what it means to be a specialist, and also something to do with NLP. So can you maybe give us a quick tour of what you were doing before you started? Your own company?

Oksana Tkach 

Yeah, well, localization specialist is basically what I call a translator, or at that time, what I call the translator, because it just sounds fancier opik. But also like translation and localization are kind of different than localization specialists has to do all of this stuff with like translation memories and cad tools. So I just kind of I chose that title. Yeah, and I was doing that because I was learning to program. And but also, I kind of liked the localization field, it was kind of exciting to me. So it was just kind of translating, and have found, so worked for blablacar. And the company, the ride sharing company in Europe, I was really cool gig because the team was just really great. And the content was really fun. And it was a really big challenge because it was like creative content, right? Like even the interface had to be creative, like snappy, because Google is very sort of strict. And that was like, my brain was just melting. Just I was trying to force myself. I mean, it would always come out fine at the end. But in the process, I was like, I have no idea how to make this text more engaging. But in the other thing, I just kind of learned some writing skills during that job as well. And then yeah, and when I was working for them. During that time, I started to get NLP jobs. My first NLP jobs because kind of the market was growing a little bit.

Andrej Zito 

Did you? Did you look for the jobs on your own? Or did you just fall into your lap somehow?

Oksana Tkach 

No, nothing fell into my lap at that time. It was Yeah, it was brutal. Because I was feeling I was feeling really weird, because I was learning all of these really advanced, complicated stuff. But then there was I didn’t feel like there was demand. But and also, everybody was talking about AI and data science and blah, blah, blah, and natural language processing. But there were no good projects. But then, it also was super intimidated, because that’s where a neural network started to become popular. But I was just only started, Tino was starting to understand how programming works and how it applies to language. But then, but now it’s like you have to learn but neural networks know this advanced calculus stuff. So I would like go to a workshop. And it would be at a workshop like in your electronics workshop with like these programmers, like four year old programmers. And they were so much faster than me. And I was like, Oh, my God, what am I doing here? Yeah, but eventually, I just, I started to get tiny jobs doing computational linguistics, and then kind of,

Andrej Zito 

So, so okay, maybe it’s not the right time. But what does it mean getting a tiny, computational linguistic jobs and what exactly those jobs look like? Explain it to us. Okay, mortals, because we just know like translation localization, but computation linguistics, that’s something super foreign to me. So what what exactly is expected of you.

Oksana Tkach 

So one of the first jobs that I got, we were building this engine, so this huge app that would look at the text and find automatically find named entities, so like names of countries, people, addresses whatever. And but it was supposed to work for like 16 different languages, and that included like Turkish and Finnish. And so one of my tasks was to evaluate. So we had this thing called lemma tiser. c would take a limit. Iser is an algorithm that takes any word in any form, and it tries to convert it to like the original dictionary form, you know, like, if it gets the word dogs, then it has to convert it to dog like, it has to note that as is just an ending. So in my job was to take the results of the finished liquidizer and check it and I was like, but I don’t speak Finnish. And so I had to like go and figure out how the nouns are like how the forms of the nouns are kind of changed in the Finnish language, which is so weird, like they have so many different forms, like they don’t, they don’t use prepositions because instead of preposition, they would just like put a different form of the noun.

Andrej Zito 

So how did you approach that problem? Like, did you go and learn the Finnish grammar or?

Oksana Tkach 

Yeah, you go and read about Finnish grammar, and then you and then you did use a google translate do, or like you go and look at the dictionary and see, okay, this word probably means house. But then in the dictionary, it has this form. But amortizer showed a different form. So it’s probably a mistake. So that was a challenge. So there is something called. Yes, that is that was like a perfect competition linguistics project example.

Andrej Zito 

Can you give us a context of like, what was the real application of that project that you were doing it for? Like, why would they need to do to grab the the nouns?

Oksana Tkach 

Oh, no. Okay, I’m not gonna disclose what exactly that company was doing.

Andrej Zito 

That’s okay.

Oksana Tkach 

Yeah, but like, for instance, if you have a chatbot, so an intelligent, like conversational agent, and it can book flights or something, and you can text it Hey, bot, I need a flight from like New York to wherever to cave, it’ll automatically identify the names of the cities. And then if you texted the date, it will also automatically identify the date. And then it’ll use this information to send the request to a website with the with flights, and you can just like book the flight for you. So it’s something called natural language understanding. Understanding is so it’s not just set, like a set of characters anymore, it actually kind of extract some useful information from a sentence, right?

Andrej Zito 

Yeah. Oh, yeah. I was just reading about NLP yesterday on Wikipedia. So the just the, the rough surface? Yeah, there was the the natural language understanding and then rotor natural language, is it creations generation generation.

Oksana Tkach 

So when you hear, uh, you know, the news about, like, scientists invented in neural network that generates fake news. So they basically trained, they trained a neural network on a bunch of real articles. And in the process of training, it just kind of looks at the frequencies of words and how the words combine. And, like what which words appear in the beginning of the sentence and the end, and the structure of the sentence, and that it can kind of like, tries to generate something that corresponds to the rules of our actual language. So it reads like it was written by a human, but it’s actually just kind of like numbers.

Andrej Zito 

Another thing that I learned yesterday was the statistical learning, kind of like, based on the rule sets, right? And then the machine learning. So like, in your first projects, were you creating the rules? Or was it also already based on machine learning?

Oksana Tkach 

Yes, that was a rule based project. So basically, what we did they later on, we would also like, right, okay, if, if you have the words like company followed by a capitalized word, then that capitalized word is probably the name of the company. So there is something called rule based NLP, because you get a bunch of linguists that are really good at understanding language, the structure language, and then they write down really like formal definitions. Like they describe patterns. So we see in the language of the patterns that we use, to say, the name of the company. So just kind of like, we are using this sort of a subjective language to select objective knowledge of the language to kind of describe it. And then the other method is machine learning, in which we take a lot of data a bunch of raw data, and works really well if it’s labeled. So if somebody like a human, went and like highlighted all of the names of the companies, then we can use machine learning to objectively look at the context that we see the words that we see. Before and after the company name, and kind, and then we kind of let an algorithm define the rules inside even the rules. It’s kind of connections and patterns and numbers, but it’s not done by human it’s done by a computer that just like does math on just on the, on the words that it sees in the data set. So those are like the two major approaches to language.

Andrej Zito 

The example that you gave us, you said it was like a first tiny job. So how did the later jobs that were like more complicated look like?

Oksana Tkach 

Okay, so I worked at this startup, and I was creating a bunch of chatbots for them, like super intelligent. I mean, to me, because it’s my work. It was like that. That was, I mean, that was super intelligent, because other templates that I see on the market right now are not that intelligent. Yeah, so it was basically a childhood. Trying to use pretended to be a person and like you weren’t supposed to understand that it’s a, it’s a chatbot. And then I did a bunch of neural networks for this other company. And then I was just kind of. So this was about two years ago. And I just got to hit a wall. I just didn’t like the projects that I was getting. And I was just kind of dissatisfied with how everything was going. And I kind of missed translation as well, because I had left that field for for a few years. And then somebody approached me, and he said, they needed localization engineering. And I decided, Okay, I’m just going to open up a small agency and hire some people on because I also really wanted to learn about localization engineering, and how all the technical side of translation works. And so I decided, Okay, whatever, I’m just going to take all of my savings and open this and open the staining company. And that was a year and a half ago. And right now. So right now we, basically, we split, so we have some people doing the position engineering and some people already doing and it’ll be,

Andrej Zito 

Yeah. So before we, before we get to Metamova, I’m still curious about the chatbots. Because I did work on localization of one chatbot for one of our big, big customers. And my very amateur question about chatbot is, let’s say you have a chat bot that helps to sell I don’t know, clothes. right, you can chat with the bot, and maybe you can make a purchase with the bot, right? And then let’s say you have a different company, which sells I don’t know, they sell protein powder or protein shakes or whatever. So would you that So my first question is, do these different companies need to have like their own chatbot created from scratch? Or is there something like a commercial chat bot, which helps with e commerce in general?

Oksana Tkach 

Okay, so the problem with chatbots, especially with Ecommerce chatbots, is that they have to have very narrow context to work properly. Which means that for each sort of task, you would, so you would basically, for each interaction, you would need to, you will need a programmer or an NLP engineer to sit down and sort of reprogram it. So if your interaction is the same, like a greeting, a greeting would be the same for everybody. But if somebody wants to order protein powder, as opposed to clothes, you would have to go in and completely redo that particular interaction. So big. So basically, you would just need to start a new project for each new chatbot. And I mean, there are ways to sort of make it faster and scale it a little bit, but I don’t really, I had when I was working on chatbots, I had this idea, why aren’t we exchanging interactions, right? Like, why can I take somebody else’s programmed greeting thing, and just use it for me. But that’s what kind of what Amazon does with Alexa and Google with Google Assistant. Because they let people code for different kinds of interactions. But then they don’t really share it. So they just heard it, they have this enormous data set, but it’s just this enormous program that kind of knows all of the phrases that the person can say. So they are building this giant, all powerful chat bot, but you don’t really have you can only contribute, you don’t really have access to other people’s code.

Andrej Zito 

Right?

Oksana Tkach 

If that makes sense.

Andrej Zito 

Yes, it does make sense. That’s, that’s basically what I was asking about, because I’ll give you an example. And again, it’s something that I’m not totally knowledgeable about when you talk about empty engines, right? So you have like the empty engines from Microsoft from Google, right? So you can like, use them as like your baseline, and then you just feed them your own data, if I understand it correctly. So is there something like that for chatbots? I think like, that’s what you were trying to point out. Like, there’s something like, like a big, like well trained, very smart chatbot that you can use as a baseline and then just kind of like, optimize it for like your own contexts or for your own needs.

Oksana Tkach 

Okay, so the problem is that chatbots don’t really work on neural networks, yet. There are some efforts to do that. But the problem is chatbots that are trained in neural networks, it’s very hard to predict the response. So you can you can create something that kind of simulates a conversation. But if you need a chatbot, to say, Okay, here are all of the shirts and the prices and the colors that we have, that has to be hard coded, because the neural network, or at least the neural networks that we have, right now, you can really guarantee that that’s what it’s going to respond with. So machine translation is different, because you just kind of throw a bunch of sentences. And it kind of gauges the how the language works in relation to the, you know, to the target language. But with chatbots, when you have that really specific task of selling of being basically an online store, but to a chat, you need to really hard code, a lot of stuff. So when I say a super intelligent mega chat, but I say that it’s, it’s complex, it has a lot of parts, but the parts themselves are pretty simple. Right? So it’s not being trained, it’s just being expanded. But machine translation works a little bit differently.

Andrej Zito 

So then, let’s get to your final stage. Where are you now? Right? CEO Metamova. saying the word final stage when there’s Coronavirus, probably not the best idea. But let’s say the current state current stage of your life. So you mentioned that you got there because somebody asked you to do an engineering job. So I’m wondering how they knew about you, because you didn’t mention engineering before? So that’s the first question and why did you think that you actually need to create a company around that?

Oksana Tkach 

Oh, that was okay. That was a really weird kind of synchronicity moment. So I was at that time, I was living in this feared studio. And the floor was like, full of kind of students and people my age like young people. And we all knew each other. And this one guy, he had kind of like a mentor. Somewhere in I don’t remember in what country like Stockholm, or something, some Scandinavian country. So my friend’s mentor guy met the guy who will become my client. And yeah, and they just kind of passed that information to Ukraine. And so my friend was like, do you know what localization engineering is? Because I need to find some people. And I was like, Yes, I was born to do this. And yeah, and he just connected, connected us. And I sent my resume. And it just said, that I had experience in translation, and in programming, and in data science, so that was kind of nice. And I just decided, I basically so the guy, my client, he offered that I could either do some just do some work for them. Or you could have a team and kind of take on more work. And I decided that I wanted to see if I could do it. I wanted to see if I could like dare to hire some people and say, you know, work for me, which was really weird. Yeah, so it’s just like, Yeah, of course. I think it would be a great bus. Yeah, I just did it.

Andrej Zito 

Okay. How did you pick the name? What does it even mean? Does it have any?

Oksana Tkach 

It means meta language in Ukraine.

Andrej Zito 

Meta is sorry. Mova is language?

Oksana Tkach 

Yes, mova is language in Ukrainian. It was just like, well, because I was thinking, Okay, if I’m going to be doing NLP, eventually, right. And the key is, you’re creating, you’re using a programming language to describe a language. So meta seemed appropriate. But it also had, like, nice sounds, do it like em and open. Ah, that’s, that was really kind of nice and feminine. And if you look at my logo, it’s pink. My website is pink, which was, and when I started, when I hear that word, I kind of see pink, which is kind of really nice. And it’s also a little bit political, because we are always having this, you know, thing in Ukraine about the Russian and Ukrainian language. And when the Russians try to like make fun of people speaking Ukrainian, they will use the word MOBA. So it’s like, I’m gonna I’m gonna add some political stuff too much to the name of my company.

Andrej Zito 

I like it. I like it. There’s so many subtle messages to the world. Yeah, I checked your website and I and I really liked that the sign. It’s a one pager.

Oksana Tkach 

Thank you. I did it myself.

Andrej Zito 

Oh, nice

Oksana Tkach 

The whole thing.

Andrej Zito 

Awesome. Yeah, I think

Oksana Tkach 

it’s so much fun hosting and stuff like,

Andrej Zito 

yeah, so that’s another thing. Let’s let’s talk about that. So like he decided, Okay, like, I’m going to create my coffee. And then I’m just going to hire people and, and run the company. So did you also learn something about it on your own? Did you just improvise and learn on the way?

Oksana Tkach 

I mean, no, I just I like to say it was a dumpster fire. It was like, Oh my god, where am I going with this? What is going on? But yeah, I just kind of made a bunch of random decisions. And some of them worked. And some of them didn’t. I just, I had a goal that we were going to do this for a year. And so no matter what happens, we have to hold on for a year. Even if like, I don’t like the work, I don’t like the client, like a bunch of people quit. I hired new people, it was like this whole thing. But I just always had this thing. I’m doing this for a year, it was really weird. Because in a way you can, if it’s your thing, you can drop it at any time, right? But if because nobody is making you do this, but if you promise this to yourself, it almost becomes kind of like a baby. Because like you have, you know, when you have a baby, you’re like, you can go anywhere, like you’re stuck. Even if it sucks, you have to take care of the baby. To me, it was like, Well, no, it’s my baby. And I just have to push through it. Yeah, there was there were times where I just didn’t have any money for a couple of months, because, you know, that’s how business works. And so the year, so we moved in to into a real office in September of 2018. And so this September of 2019, I kind of counted that as year. And I was like, Okay, now I can do whatever I want. And it was so much more fun. Because now I just kind of, first of all, I know that I can do this. And second, I don’t have I don’t have any, like obligations before my, for my to myself, I guess anymore. So just kind of do whatever I want. It’s much more fun.

Andrej Zito 

So, how are the early days of the company, did you just have that one client? And you have to figure out what localization engineering actually means?

Oksana Tkach 

Yeah, yeah. Oh, my God. Yeah. So basically, we started doing work, I think in June of 2018. So from June, to September, I was just going to be learning really fast. But all of these because we got a bunch of accounts, and all of them had different types of processes. And so I had to hire people make sure they got understood what was happening, make sure they have like internet connection, computers, and then also learn about localization engineering. And I traveled to where my client was, I stayed there for a few weeks. Yeah, and then I think around like November, I was already kind of bored. Because I was like, oh, okay, I know. I know. Yeah. Engineering guess. Yeah. So that it was just kind of like about business, a business was figuring out management and stuff like that was much harder for me than localization engineer.

Andrej Zito 

You mentioned that you were doing a couple of accounts. So does it mean you got this couple of accounts through that one client?

Oksana Tkach 

Yeah. So the client, I mean, they could give us more work, right? They could kind of cover my whole team. But I didn’t feel comfortable having just one client. But it actually took me quite a while to kind of get brave enough to find new clients. Because that’s kind of hard. Like if you want to expand like, we need to hire someone new and pay them. But maybe the new client hasn’t come in yet. And you didn’t have the work to give them like that’s much more complicated than just having one client. So that took me a long time. But right now we’re kind of like, it’s much more balanced, because we have several clients, and two different directions that we are going in.

Andrej Zito 

So I saw this on your website. And now it kind of like makes sense. Because you started with engineering. with Kenny, you maybe explain to me why somebody would just outsource localization engineering, versus going to like, let’s say vendors that can do the whole localization in one package.

Oksana Tkach 

Because this was a vendor. So they weren’t doing the whole localization. But they were looking to kind of get more engineers. And I guess the place where they were There weren’t enough people. So they tried to outsource to Ukraine. So Ukraine is a popular place. And to me, it was a great opportunity. So I just took it,

Andrej Zito 

When you started the company was your kind of like a vision from the start that you will eventually do something about NLP. And like engineering is just like, okay, like the first step to to get off the ground.

Oksana Tkach 

Okay, so my even my initial plan because also for this client, so we do engineering, but we also do a lot of development. And my initial plan was like, okay, because if I’m doing development, they will eventually need an LP because it’s a translation company. And I was so wrong. Because newsflash, large vendors in translation, are doing an LP, which I think is a huge mistake. But that’s just kind of how the business works. And so we pushed it, but then it kind of explained why they needed to do that. But it was just not. It’s hard to get that kind of innovation implemented in a large company. So I had also hoped that I would find, like, other smaller vendors, right, and do an LP for them. But I don’t think there’s space for that in the market. I don’t think like smaller, maybe agencies do actual an OB. So they want to, like Connect existing machine translation engines, but they don’t really want to do like cost them an LP, which I think I thought would be really useful.

Andrej Zito 

Yeah. Okay. So this is something that we already talked about when we had our first chat. And I come from the probably the mindset that localization does localization, and we don’t know much about NLP. But for you, to me, it seems like for you is like very obvious that we need NLP can you actually explain to us like, what do you mean that like, translation vendor would use custom NLP? Like, in which situations?

Oksana Tkach 

Yeah, I did, like I did a webinar about this once. So there’s so many different tiny things that you don’t even you just think it’s development, but it’s actually an op, like, every transition process needs text pre processing. So you have to split the text into sentences, you have to segment it. And then, you know, when you do QA, you have all of these plugins that do QA, how do they work, they look for terms in the source segment, and then the corresponding term in the target segment. And that’s really close to what I was doing in that NOP project with finding names of countries, right. So all of that automated quality assessment is basically NLP. It’s just done by developers that I met that are making cattles. And it’s very often inferior, because they don’t really understand how language on the structure of language like they don’t know what’s in the middle. So there are a lot of false positives and a lot of these like, useless tips that the plugins give you. When we got so many other things, like when because when you work in localization engineering, right, you see all these problems, like, okay, you get text, there are some icons, now you have two icons, one convert to this other thing, they have to find a way to do something with icons, but that’s also taxed. So I use when I do development, I use NLP algorithms to handle stuff like that. To do localization engineering makes sense? No. Sometimes when you have different cad tools, and for instance, like you don’t want to pay for products, because they’re super expensive, but your client is giving you a translation memory from others, but the translation memory has different segmentation than your traditional memory. So they are conflicting, how how do you leverage that translation memory? Well, we built this algorithm that splits the kind of intelligently splits the original segments into sentences because that segmentation was by paragraph, right. So the segment was a paragraph. But we needed a segment per sentence to basically converting try the stem into a different format of template. We’re using an NLP algorithm to split just some things called so is voluntary disambiguation there are a lot of tricks you can do to kind of save money to leverage the stuff that you have, by using NLP,

Andrej Zito 

didn’t you think of contacting, like, SDL, or MemoQ or Memsource? and kind of like help them improve their cat tools?

Oksana Tkach 

Um, I mean, I’ve met the representatives at conferences. But to me, it’s like, I didn’t think they would care enough because I didn’t think they would invest enough resources for me to actually for me to, to make for it to make sense to me to improve like their plugins, I didn’t think I didn’t think they kind of, they would value that as much. If something kind of works, something works 70% of the time. If the stuff that their developers wrote works, then I didn’t think they need Oksana to come and save everybody. Because I used to think I’m gonna come and save everybody.

Andrej Zito 

Else, I also have the hero complex.

Oksana Tkach 

No, I have the, this is wrong. It shouldn’t be like this complex, but then everybody’s like, we have budget.

Andrej Zito 

And budgets. Yeah. Yeah.

Oksana Tkach 

Okay. Sometimes it makes more sense to use the old things in place. And then to try and disrupt everything. Sometimes disrupting is kind of more, is more expensive to accompany. So I just got to I gave up on trying to reason or make friends with companies like that. Okay, so my kind of global thing is, I like to think about how artificial intelligence will work in the future, right. And I think it’s the way we have a trade now. It’s not going to produce actual good AI, right? I believe we have to sort of make the economy more distributed for that to happen. And make data more not as make data aggregation. More up to the owners of the data, right. But the way I started to think about this was it was in the context of the economy of translation. So my idea is I’m going to make this kind of small example of how I think AI should look. And I’m going to make it on the basis of that sort of reinventing the translation process.

Andrej Zito 

Okay, we’re, we’re all excited to hear about it. Please give us some details.

Oksana Tkach 

Oh, my God. Okay. Well, okay, so I believe that the translators should be picking the content they translate, and also the actual users of the content, right? So the content that people want to see translated, should be more expensive to translate. That makes sense. Because right now, what we see is, companies have money, and they have this like, website that nobody really wants to see. And they pay to translate that website. But like, what is the use? Okay, so we’ve expanded the kind of the content of one particular language, but not in the way that the users of the language want to be. I think useful stuff should be translated like Wikipedia, Coursera, whatever, and non academic articles, whatever gives you more language or whatever, it gives you more knowledge. Because right now all of the knowledge is in English. And it’s all centralized in English, right? That doesn’t make much sense to me. But the way you can do it, is to kind of give the translators the ownership of the of the translation memories, and let them decide what to do with that. Because the big drawback that we have right now, is all the translation memories that we generate. They are under NDA, and they all are on lockdown by vendors, because the clients that sent the text for translation, they don’t really care about the recession, memories. They just want the stuff translated. And sometimes it needs to be released, like in a month, okay, whatever. But they don’t really care about this bilingual data. You know, who really cares about the bilingual data though Session memories look, his data scientists, because you don’t have enough data to train really good machine translation.

Andrej Zito 

I’m thinking so many things. So I’m I’m doing my analyzes paralyzes where I’m just silent. And I don’t know which question to ask.

Oksana Tkach 

So kind of like, okay, I’ve said something. Yeah.

Andrej Zito 

So clients on the on the things, right? Because they’re paying for it essentially, right? Yes. And I’m not sure, like if it’s like in all cases, but I’m pretty sure that at least some of them are using the translation memories to train their MT engines, do you think it’s an exception?

Oksana Tkach 

Okay. So they probably, here’s how it works from the point of view of data science, they probably use them, whereas statistical machine translation was popular, because the second machine translation needs way less data to kind of start working results start showing results, then come huge companies like Microsoft, they have their own machine translation. But that is because they have so much. So many translation memories, that they can actually train a neural network, because a neural network that produces much better machine translation results, needs a lot of a lot of a lot of data. So probably companies like Microsoft do help their customers the engine. But smaller clients, they don’t have enough data to actually train their own custom. And the one solution that exists, already exists, but I’m not sure how many companies use it is tuning. So if you know something called auto ml, by Google, you can take Google’s empty engine as a base, and then you can upload your translation memories. And basically what happens is, they kind of unzip the empty engine model file. And they, they kind of move the weights of the words in the direction of your translation memory. So what happens is the output is kind of still superintelligent the way, empty the way Google’s empty engineers. But it kind of favors the phrases and words that it saw in your in your transition memory. I don’t think a lot of people use it. Because to train to make this tuning thing, you need to upload the the memories to Google, and they probably keep them because Google is interested in building a huge data set of traditional models. And I don’t think a lot of clients allow that. So what happens is at the end, it’s not you can do this, but it’s not really economically viable. And what vendors or clients or whatever and doing is they just pay for empty API. Like they pay Google to use the baseline, basic empty engine. And they translate, and then they give the job for post processing. And sometimes it’s good, sometimes it’s not good at all. But just the mere fact that they are paying for post process post editing. And that translation allows them to lower the rates dramatically. And so that is the big injustice that I see right now. The rates are dropping, which means that there are more jobs, because now more more clients can afford to translate. But then the translators, I don’t think the amount of work that they do right now in an hour is justified by the law race that they are getting paid for posts. So the post editing could be easier and much, much faster. But it’s not. But we still have this

Andrej Zito 

It could be easier. If If what happened.

Oksana Tkach 

We started to tune the machine translation according to the client’s data, if we started to tune it according to the translator style, because very often the translators, as like artsy people, they make corrections just based on Well, I prefer this phrase, I think this phrase is grammatically correct. But I think it’s like, whatever doesn’t sound good. So that takes time as well. But what if the machine translation was almost the way you would do it, but you are just now correcting like a missed word, or the wrong pronoun or something. But that is kind of right now. It’s, it’s very hard to unlock it because we have Google that is kind of a monopoly and they have the ownership of this enormous engine. And then we have vendors that are home ownership of all of these transition memories. And then we have the translators that could make could just, I don’t think they could make more money per word. I think they could do more work per hour and make more money like that. But it’s not happening. So I think there is a big imbalance happening right now. And that is because we have all of this innovation. But I don’t think our industry is adapting fast enough to the innovation that is happening, because it’s so kind of in the 90s. It’s kind of 10 years ago, but not the 90s, but 10 years ago, and it’s all kind of bloated, you know, that it’s very hard to run when you have so many kind of uncomfortable things strapped to you.

Andrej Zito 

Mm hmm.

Oksana Tkach 

If that makes sense.

Andrej Zito 

Yeah, yeah, he does. Yeah. So you mentioned that like, the word tuning, right. Did I get it? Right? Okay, tuning. So from what I understood, like the the main problem that you see with the tuning is that if, let’s say like, some small high tech company, let’s say in Silicon Valley, I’m a startup. And I do some translations. And if I want to use the tuning process, then the main risk for you is, is that if I want to use it, I have to basically give up my TMS to Google, right?

Oksana Tkach 

I think so. I, because the only way, because the other way would be to download Google’s engine file, and they are not going to let you do that. The other way is you upload it to Google and there is no way that they are not keeping that data, because they are letting you use this thing for free. So the you know, the pay is that you’re giving up the data.

Andrej Zito 

But what is what is the what is the danger for me if I give up the data, like we all give up the data in this age that we’re living in, right? Constantly.

Oksana Tkach 

I mean, there is no danger, danger, I would encourage to share data as much as possible. It’s, again, it’s the NDA. And also, you’re making Google more powerful. So you’re making Google more of a monopoly, and also Google, which means Google is free to raise the Machine Translation API prices. However they want.

Andrej Zito 

Oh, it’s not like a free thing?

Oksana Tkach 

I think so it’s like for amateur use, it’s free. But then if you want to translate something like more than 1000 words, it’s Yeah, it’s getting expensive.

Andrej Zito 

I see. So what do you think is like the the alternative to this?

Oksana Tkach 

Yeah, I think we should just all share the data in some secure way. And we should have a sort of a communal machine translation engine, which kind of belongs to no one as well, which, and this idea was inspired in all disclosure by blockchain. Um, because that was like the first thing that I’ve heard about that offered like a third alternative, right? Because what what else do we have, we either have nothing or we have a monopoly. But now we have this third thing, or at least the idea, the concept of a third thing, where we could kind of all have equal ownership of the empty engine. But then we could also all equally contribute to it. So whatever transition job you are doing, it’s updating the MT engine, but you’re also getting paid for updating the MT engine. And it improves every day. But okay, and this is an important point to make. People will argue that well, MT engine, neural network results, they stop getting exponentially better. Once you’ve used up like, you know, 100, gigabytes or whatever, of data. But my point as a linguist is that the way we use language changes every second, we come up with new words. There are new means. There are new stuff that happens there are new viruses forever new breaking news. Which means that if you train your mt, a week ago, and you stop training, that it’s an you want to want to use it to translate something really important that you need to know right now. It doesn’t know what’s going on. So you have to always update it. It doesn’t. At some point, it stops being about language quality, it becomes about language use, right? It needs to be up to date, always. And that’s how I think AI and neural networks should work in all fields. They have to be updated. Second, something new appears online. But the only way to achieve that the only way to make the data sharing safe. And the ownership safe is to distribute it distribute the ownership. Because otherwise, the other option is give up all of your data to Facebook or Google or not, don’t show your data at all. Which is I don’t think that’s the answer, because then we are stopping the innovation. And what’s the point of that?

Andrej Zito 

How would How would? So let’s talk a little bit. I like the idea. Like, I like the vision. How will it work in practice? Let’s say how do I know that like the people that contribute to making the empty better are actually making it better? Because you know, from your own experience, there are many bad translations out there. And what is good for me or bad for me may not be good or bad for some other people.

Oksana Tkach 

Okay, so this is the problem of translation assessment, right language quality assessment. And I, when I was a translator, this was so annoying to me, because the reviewers that would review my work, I would it would be one person and the errors, they would find they were very often not in there at all. So it’s very much subjective. And I’ve been thinking about how to make this more objective. And for a while I thought, well, maybe I will create a neural network that corrects that language. But which, which, by the way, something that memsource has done recently, but they they created a neural network that assesses empty results. But as I was learning more and more about how neural networks work, I realized that you would, you would use the exact same algorithm as you would use for machine translation. So instead of creating something that improves language, you might it’s the same process as creating something that translates perfectly from the standpoint of data science. There is no really, there is a difference, like if you want to, the only way to know if a translation is the only way to let an algorithm know if a translation is good, is to create an algorithm that translates perfectly. That is the big problem. So that’s why I’m really skeptical about using neural networks and stuff like that for quality assessment. And so what that lets me realize is that, but then we need humans to assess language quality in any way. But how do you make that objective? And that also, this idea also comes from blockchain? Why not make it a consensus? So if you say, randomness is a big thing in blockchain, because randomness ensures security. If you don’t know what happens next, it’s impossible to hack a system. So if you don’t know who is going to be assessing your translation, it’s impossible, there is no incentive to submit a bad translation. The only way you get paid for translation is if other translators think it’s good. But if you don’t know who the translators are, it’s impossible to conspire with them. And there are so many problems with this idea. I know because if you have very, like a very small pool of translators, it’s very easy to consume in blockchain in like decentralization, the bigger the network, the much more secure it is, because then the randomness is very hard to predict. But yeah, so my kind of utopian idea. And by the way, who knows, is going to work, you know, Bitcoin is kind of working, but isn’t working, I don’t know. But this is kind of what I spent my days thinking about. If If we let let’s say, five translators anonymously, evaluate your translation. But then we only pay to the people that created or voted for the good translation, then there, the incentives are perfectly aligned. Because it’s completely objective. And everybody, it’s in everybody’s best interest to create a good quality translation.

Andrej Zito 

You were mentioning this in the in the first call that we had, and I didn’t say something, but it came to my mind. So let me say now, in one of the earlier episodes when I was still doing this on my own when I was just going through like the ladies articles, like on social media about localization and I was just talking about it. One of the articles that I covered was that how Facebook is evaluating machine translation quality and it kind of like the To what you were saying before that we should be translating what the people would users actually want. So why should the quality be assessed by the translators? why shouldn’t it be assessed by users? So what Facebook does is that they feed to different people, they basically a B test different empty outputs. And based on the engagements that the users have with his machine translation, they kind of like evaluate like, which one is better?

Oksana Tkach 

Yeah, I guess it would work, too. I guess I was, I was I had been thinking more globally is, in a way that if you have a transaction on any website, and you don’t like it, you can report it and say, you know, this is not I don’t understand what’s going on here. But I mean, it’s kind of hard. It’s not going to work for everything. But yeah, you got to start somewhere. Yeah, I mean, I like that system. If Facebook does, but the thermos intersession sucks. I’m not sure. I’m not sure if it’s working well. But actually, Facebook. So Facebook already has this really cool thing. They have translators community. So do you know that the interface of Facebook is all translated for free into all languages, because they have these communities, and have this special app inside Facebook, where it’s very similar to consensus, except it’s not anonymous. So you get, you get the English phrase, and then somebody translates it into like, let’s say Ukrainian, and then other members of the community vote or they propose a new translation. And then whichever version wins, that gets posted as the translation of Yeah. So I kind of I kind of took that idea from Facebook a little bit. But the problem about that is that it’s crowd sourced. And it’s not, I mean, it’s free. So they’re just basically trying to get content. Another problem with that, is that the data that they have from people translating for free, it’s very hard to apply that to actual posts that people post, because that field is web, it’s web interface. But then if they are trying to machine to incite people’s posts, it’s completely different style. suits are going to help them.

Andrej Zito 

Yeah, it’s UI versus user content.

Oksana Tkach 

Yeah. And also huge problem right now. Because to me, I mean, interface is important. But interface is very often intuitive, especially on our phones. So it’s kind of easy, even if you don’t know the language is easy to understand where to press, for me more pressing is the problem of like translating Facebook posts, or tweets, or, like stuff that people actually create? Because, like, so many news, nowadays are shut or shared through Twitter, like what is going on in China? You know, like the protests in China, who knows? It’s all in TV. So, but the problem is that all of the empty engines are trained on like, very, very nice articles that somebody took from the European Union, or, you know, the law articles or whatever. Because there is no bilingual data for everyday language like that, like, how do you? How do you teach an engine to translate memes? Like that, that’s we need to there is no data set like that. So we need to generate a data set like that somehow.

Andrej Zito 

somehow. Okay. So we talked about the big picture, and you mentioned to me that you already have something in progress. So can we talk a little bit about your solution, or like how you’re going to test your assumptions?

Oksana Tkach 

Don’t get-Okay, so I’m just trying to until you build like, a kind of a minimum viable product example of how I think they should look. And, to me, the perfect source of the content that I want to use for this sample project is Twitter. So my idea is, how do we translate tweets effectively, right? Because Twitter is full of memes. It’s very specific language. And I found out recently that until 2017, Twitter was also trying to let people translate tweets, but they closed it down. I think it wasn’t great results. Like I didn’t think I think people were just translating whatever or were interest hating at all. But I also I chose Twitter because it’s a great example of a network right? Because you can see tweets are either not popular at all, or super, super popular. See, I either get five likes, or you get 20,000 likes. And so I am interested in taking those super, super popular posts. And also Twitter is interesting in the sense that if you don’t follow somebody on Facebook, you’ll never see their content. But on Twitter, if you don’t follow someone, the repost mechanism works in a way that if a tweet is popular, it will eventually reach everybody in the network. That’s why Twitter is so cool. So my idea was, but also because Twitter has really great API, so it’s super easy to get the latest popular tweet. Some idea was, okay, let’s get like the last latest popular tweets. And let people give people a way to go and sort of contribute and let their translations be evaluated by consensus. And then immediately when something is confirmed, we’ll kind of like post it on Twitter, but as a repost, you know, so you’ll have a tweet, and then you’ll have repost of that tweet, but in different languages, kind of like a social experiment, whether that would work or not. Because my, my big question is, would people care enough. But the more popular tweet is, I feel like the more people care to see it in their own language. So yeah. So the back end is kind of more or less done, I’m now going to working on putting it together and making it usable. And then, and then we’ll see if it’s if it’s ever going to work, but I would just like I would like to put it out there and see to see how people react. And the idea is, as more people translate, the more data we have. And so we can train our own engine, that translates tweets. So the easier it is to translate tweets, so the more people translate, and the faster people translate, so the more data we have. And also, the idea is that because we have this engine, we can now use it. We can now you know, sell it to other social media platforms, the use of it, and then use that money to pay the translators. And so the more you translate, and the battery translate, the more money you make. So that is really kind of like full circle. digital currency. Translation coin type of a thing.

Andrej Zito 

So are you planning to have like some third party app? That would use Twitter’s API? Or are you thinking of somehow integrating it into Twitter? I’m not even sure if it’s possible.

Oksana Tkach 

Um, yeah, you can have. So a couple of ideas that I had, we can have a separate website, we can have, like a plugin that you install on your browser. If Twitter is open to it, they could just add a button to Twitter, but they you know about me yet. So. But that’s just a matter of time. I mean, I didn’t, it’s not impossible there. There are a couple of ways to go. But that is also a big question that I have, whether people would even like go to my website, a separate website and translate all this stuff. But you know, a lot of people would say, that is a horrible idea. And it’s not going to work. A lot of people would say, that’s great. You’ll have so much data, blah, blah, blah. But I don’t trust any of those. Like I don’t I don’t even ask people about because I used to pitch the idea to everybody. And then it was like, I did not hear any valuable advice. And so I just kind of stopped. Because that’s that’s the problem with new ideas is that it’s new, and so nobody, nobody is in your head. And nobody really fully understands it. So there is no point in asking about how to go about it. So you just kind of do it. And you see what happens next.

Andrej Zito 

Do you have any timeline? For your MVP?

Oksana Tkach 

Well, the Okay, so I had a plan before New Years I worked out a plan. And the first quarter of 2020. I was I was supposed to finish my pitch deck, and it’s not finished. So I have two weeks left, maybe I will Yeah, maybe I will finish it because I have nowhere to go because it’s Coronavirus outside. So we’ll see maybe maybe I’ll postpone it until May. But But I just got it. I wasn’t planning to build the actual app yet, but I just kind of got inspired. And I just did it in one night. So now I can’t

Andrej Zito 

Wait. So you had the app, you’re just stuck on the pitch deck?

Oksana Tkach 

I mean, I guess. And the app needs for it as well. But it’s just it’s it’s it’s a it’s a work in progress.

Andrej Zito 

Who is the pitch deck for?

Oksana Tkach 

potential partners? potential investors? Who knows? I don’t know, again. You just do you just send it out to just, you know, let people know about you. I’d have no idea. I have no plan. That’s not a great fit for me. But I mean, I have different ways it could go. But yeah, I’m not. I am humble enough to know that neither of them will probably be exactly as it happens. Yeah, so timeline, I didn’t know. April, to have something presented in a in a good way. Yeah, so if somebody wants to partner with me, that’d be great. If you want to steal the idea and do it, that’s also great. Because I would be so happy to see like, because I I’m not sure how to do it. myself. And if you figure it out before me, I’ll be like, cool. Let’s, uh, let’s connect.

Andrej Zito 

If you if you already have the app, did you like try to involve some translators so that they can start using it, it’s not ready for that. Alright, so

Oksana Tkach 

There is no interface. It’s all back end. So I just I know how the technology works. I’ve kind of I’ve already sort of wrote it out. But there is no like usable interface. So if you if you if you know how to send requests with Python and your translator, then you can use my app

Andrej Zito 

Is it like your solo project? Or are you working with some of your Metamova team on this?

Oksana Tkach 

Yeah, so I’ve been kind of making changes inside my team. And so I’m about to involve another person and to hire another NLP person, hopefully. And then. Yeah, and then I hope that we will have a working website. Maybe by June. We’ll see how fast we can do this. But yeah, right now, it’s kind of like, financed out of my pocket, which is painful, but I guess that’s what you have to do.

Andrej Zito 

Yeah. Okay, so this was your next five or 10 years of life. on LinkedIn, you have this interesting thing. As a description, God says, fixing the localization industry, and doing real, no nonsense NLP. Yeah, let’s start with the first part, fixing the localization industry. You already mentioned a couple of things. So what do you think is wrong? What needs fixing?

Oksana Tkach 

Oh, I mean, yeah, we already kind of discussed it, I yeah, I feel the, the way the value is distributed is completely wrong, I feel the chain is way too long. Meaning that the client base way too much. And the Crusader gets way, like not enough. And then the money goes to something super ineffective. I think that money should either go to innovation or to the translator not to do five VMs that are handling the translation. And I think the way it works right now, there are too many, too many big players that are too big to move. And so they are slowly losing money, and slowly paying the translators less and less, but still hoarding all of these resources, which makes very little sense to me. And I also feel like it’s inevitable that they will collapse somehow. I mean, I don’t know about the future. I’m not an economist, but when I look at the decisions that they are forced to make, not even are making, they’re just, they’re kind of trapped due to their size. Feels like they will be unable to go on for much longer. Yeah, and I want us to use that moment of collapse, to build something that is beneficial to the actual translators, because the translators are the people that are passionate about language, you know, they are creating something really cool. And I think they should get the value, as opposed to the management, who just wants to get more money? I don’t think wanting money should be rewarded as much as wanting to create something cool.

Andrej Zito 

Yeah, I think like when I was working on my startup long, long time ago, it was pretty much based on the same thesis, as you just said, You know, I thought like, there’s like a lot of middlemen in the whole process, that take out the margins, increase the cost for the clients. But yeah, it’s funny, because like, like, even like, when we had our first chat, when you mentioned this, like, it’s inevitable that they’re going to collapse. I always, in my mind, I see the the final scene from Fight Club, you know, all the skyscrapers exploding. But how do you actually have this experience with these big players? Did you work with them? How do you know like, how they operate? Or did their slow?

Oksana Tkach 

Um, yeah, I mean, I worked with them when I was a translator over to them when I was doing engineering. I’ve worked with all kinds of different companies, and just kind of get really closely acquainted to the work process and the, the workflows for different kinds of projects. And, yeah, it’s, I mean, and I’ve worked with them for a long time. And I came with this attitude of like, let’s, you know, let’s fix it, let’s find a little bit of a free space where we can go and do something cool. In like its startup form, so that it can be adopted elsewhere. But that did not work. And I completely understand why, like, I used to think, oh, you just stole, you know, arrogant, and you only want money. But right now, I think yeah, it just, it’s, it’s so it’s really hard to do that. In those circumstances. When you have a large company, it’s not the people personally. But it’s just that that structure ends up being toxic. In that way. I mean, toxic is a really heavy word it toxic in terms of business, not in terms of emotion,

Andrej Zito 

Did you ever have a chance to work with end clients?

Oksana Tkach 

Yeah, when I was I mean, my lablacar job. That was an endclient, because but Blablacar, it was such a cool project to work on. But the problem is that they raised a lot of money. Like they were like, one of the most successful startups in France, France. And so, and somehow, they knew that language mattered. And they insisted on not using vendors, because they were focusing on building community in each country, right? Because they were trying to engage more people to use clubs. Because they’re obviously this is a platform, you know, it works on attracting drivers and attracting passengers. So the language of the interface was really important. So they actually invested a lot of money into having on site localization managers, and possession engineers, and really nit picking the translators and building really good relationships with the translators, maybe overpaying a bit because they wanted us to be happy. And I was working really hard to produce something that was really cool. And I also had so much freedom I didn’t have to use if something was a repetition in the memory. But it didn’t match the context. I didn’t have to use it. Even it was even if it was like a context, repetition, right? If I didn’t like an authorization, I could go and change all of the transactions in the website. You know, I had access to everything, I could test it immediately. But and to them, it wasn’t a lot. But if every single end client would do that, they would lose so much money. So to me, it was really exciting. And really, it was a really creative kind of a job. But if I think okay, what if Google tried to do that? Google has so much more stuff to translate in so many more languages, that it doesn’t make sense for them to give all of that freedom to every translator. I guess it makes more sense to make it really, really strict. But that was painful to me, because a lot of the rules that Google had, in terms of, you know, the style guide, they didn’t make any sense. And then in terms of engineering, there are no anticlines for engineering, unfortunately.

Andrej Zito 

Going back to the LinkedIn statement, so the second part is doing real no nonsense NLP.

Oksana Tkach 

Yes. I love that phrase.

Andrej Zito 

So please explain. Yeah, like where do you think that NLP in other places is? No sense?

Oksana Tkach 

Well, NLP is part of data science, right? So it’s an up is just the language part of data science. It’s the language part of artificial intelligence. And data science right now is a hype. It’s a hype as the way blockchain was, maybe still is. So there are a lot of companies that really want to put AI on their website, when they don’t really need AI, or when they need a much simpler solution. But it’s just kind of really trendy to say, well, you get all of that stuff that you get from other companies. But we also put AI on top of that. And so while I was working in data science, I got so many requests from quite from clients, asking for like, ridiculous stuff that was either impossible or just not need it. So, but also, so many developers that have just background in development, but then they got into NLP later, they don’t really think about language. And they start doing something really weird. Like they know neural networks, but they don’t know about language structure. So they will start to solve the tiniest problem by training this huge neural network for a week, and then it will not perform well. To me, that is also nonsense. The way I approach NOP tasks is, I think about the field of the language, I think about language style, who is using it, why how, and then I do this kind of like linguistic data analysis. And I go and do research, in terms of like actual linguistics, like I will go and read academic articles, about like some assumptions people make about how we use different patterns in conversation, right, like the super formal linguistic stuff. And then and then based on that, I can then decide which algorithm I will use, because maybe sometimes, you don’t need a neural network to do this, maybe you just need a couple of rules written down. Maybe you need like a simpler classifier, maybe you need, I don’t know, just regular expressions. Because that is that is faster to develop cheaper, and then much faster. in production. So when an industry becomes a hype, a lot of kind of people that are selling snake oil, get into it. And then a lot of clients that really want to buy snake oil, meet them, and they do start and then they start doing this mad AI stuff. And it’s it doesn’t doesn’t work. Yeah, so I’m kind of really pragmatic in terms of that.

Andrej Zito 

Okay, so that was, that was the NLP journey, or other things that you are curious about right now. Other than LP and like your startup and your vision?

Oksana Tkach 

Curious about? I’ve tried to learn music more. So I’ve been trying to learn it for years, but I kind of have been spending more time on that. So I’m taking singing lessons, and I’m learning piano. And I would say music is about as hard to learn as data science. It’s so complex. But yeah, that’s kind of my hobby that I’m really into right now.

Andrej Zito 

Are you are you are you approaching music like from the scientific side? Or is it more like like, just like emotions and artistic expression?

Oksana Tkach 

I don’t think I am neither, like artistic or technical. I’m more the way I learn things is kind of intuitive. So I just kind of I do I start doing something and then add one, and then it somehow starts to converge. So I’ve been trying to learn piano for a long time. And it’s been really hard but and hiring a tutor was not working for me. Because the structure that she was using wasn’t working for me. Usually people that are artistic, and they’re trying to teach me something, I never get what they trying to say. But then applying logic to music is also very weird. So it’s like, okay, whatever. I’m just gonna try and learn some songs. And now, it’s kind of starting to converge all of this kind of bits of information that I’ve picked up. Now I’m like, Oh, I get what a chord is like, I get what this you know, harmony is. And so now I’m starting to really kind of starting to play faster because it makes sense how the court move That’s really interesting that standard education doesn’t really work for my brain. It’s much better for me to just kind of do whatever feels good. And then years and years down the road of like, Oh, I get it. Okay.

Andrej Zito 

Yeah. When do you actually find time to, to, to develop yourself in music? I know that when we had a call, you mentioned to me that you wake up at 1pm? Usually,

Oksana Tkach 

No, no, I wake up. I go. So I make my way to the office at 1pm.

Andrej Zito 

Which means you just opened the door, right?

Oksana Tkach 

Yeah, I just opened the door, I do not commute. So it makes it even worse. I wake up, I do yoga, I have to meditate. Otherwise, I will start to get super anxious, I need to do that. Yeah, I kind of try to play some piano, maybe, maybe like work on some white projects, which is kind of like my time when I need to be alone. Sometimes I go, go take a singing lesson, whatever. So that’s kind of the stuff that I do for myself. And then once all of that is done, I feel like I can go into work on the actual work. And usually it’s like up until midnight, but I procrastinate a lot as well. So it’s not. I mean, there are periods of time where I work really long days, but then my body kind of starts to go, Nope, that’s not working for me. And then I do nothing for weeks. So I cannot say I think I feel like people that say, Well, I wake up at 5am. And then I work 18 hours, and then I sleep You know, for four hours. It’s like, that’s that’s I don’t know, it’s a lie.

Andrej Zito 

How do you procrastinate?

Oksana Tkach 

Oh, my God, reading about Coronavirus. I don’t know. How do I procrastinate. I like to binge watch TV shows. Or I will you know i will be doing some research. And then I will stumble upon an article that has no relevance, like relevance to my research, but I will read it anyway because I’m interested. But then three hours later, I’m down this rabbit hole. Meaning about information theory. Like bits, and bytes, and Shannon, whatever noise channel, which is super complicated, but I will and I’m like, I should have used this mental energy for the other stuff. But I didn’t know. As I said at the beginning, I’m a nerd.

Andrej Zito 

I started with meditation fairly recently. I’m just using like some apps on my phone. So how do you approach meditation?

Oksana Tkach 

Do you really want to go down this? I just yeah, I have a bunch of I used to be an atheist and kind of super skeptical. And then it was making me really depressed. And so I had to discover some sort of spirituality for myself. And I have all of these different things that I listen to and do. But I feel like for in order for meditation to stick, you have to just do what whatever feels good. So if it doesn’t feel good, I don’t do it. And if it doesn’t feel good to do, like a way that I read about, I don’t do it. I just try to sit down and not think or think about something good. Or, like try to feel something that I am repressing, that it doesn’t, you know, stay inside.

Andrej Zito 

Do you do you do daily basis, like at a specific time? Or is it just only like when you find time or when you feel like you need a meditation?

Oksana Tkach 

Right now it’s by now it’s a habit. So I probably do it every morning. If I’m feeling especially if something is going on, I’m feeling really kind of stressed, I will do it twice a day. If I can’t concentrate and work I will go and meditate. Yeah, I just developed a habit. And once I have a habit, it’s very hard

Andrej Zito 

Do you use like a guided meditation or

Oksana Tkach 

Sometimes. Yeah, I don’t use guided meditations. For my morning meditations. I just kind of like set the timer for 20 minutes and I just said but sometimes I do like hypnosis, guided meditations, which are so cool. Like, because I was trying to lucid dream. I’ve never I’ve never had a lucid dream. Have you ever had a lucid dream?

Andrej Zito 

I don’t even know what it means.

Oksana Tkach 

It’s like when you are in a dream but you realize that you’re in a dream?

Andrej Zito 

Yes, yes. Of course. I

Oksana Tkach 

You’ve had that?

Andrej Zito 

Yes, yes.

Oksana Tkach 

I’ve never had that. So I try a guided meditation, but it’s kind of like you lay down and you get hypnotized. Because the person is telling you like, you can’t move now. And then you sort of kind of go to sleep. But you are fully conscious still. So that is really interesting. And it’s super long. It’s like a few hours long. But I’ve done that a few times.

Andrej Zito 

Should you actually go to like some places for that? Are there people who help you to get into that state? Or can you do it on your own,

Oksana Tkach 

Oh you can you can go to a hypnotist, but I just find something on YouTube and listen to it. But there are a bunch of different stuff you can do with a hypnotist,

Andrej Zito 

This is something that you may not be comfortable. So just let me know. I would like to ask if you could share, like maybe some failures, or just one like in your life, whether it’s professional or personal. And like how you viewed that moment now.

Oksana Tkach 

Oh, my God. I don’t look at things as failures. Because I try not to not try I just kind of I, I’ve let go of this need to charge myself to be like, why did you do that that way? And why did you say that thing? Because I’m just kind of like, yeah, whatever I’m, I’m a human. So I wouldn’t call them failures, I would say it’s like a maybe down point in your life that kind of lead you to something else. I’m just the biggest one I could think of. I mean, I was in a relationship for a long time. And then it ended and I got super, super depressed. And it lasted for two years. And it got really, really dark. And if you go to my Facebook page, I posted some stuff about that. Just because I wanted people to know that, you know, if you’re feeling really, really dark right now, I found a way to not feel that way. So that that’s a bit of hope for you. But yeah, it was like I just didn’t I didn’t want to be here anymore. Because it was too hard. But yeah, so I guess that would, I would call it like mental failure. Because my brain just didn’t want to function anymore. I didn’t want to go to get out of bed or feel anything. But I don’t think it’s a failure. I think it was a really great experience for me to have. I think it at the end of the day, it made me happier. Once I went through that, and kind of made me know who I am better. And then, I mean, I’ve during my depression, I probably failed a couple of jobs. But I don’t really judge myself for that either. This I mean, the the first year of my company, I could probably have done some stuff better. But once again, nobody could give me the right. Like, there is no person that you can ask that will tell you what to do. You know, like, you can ask yourself, that’s all you can do. So, yeah, I probably made some bad decisions, but I don’t I didn’t think I think it was I wouldn’t change anything. Yeah, I didn’t know. I try not to judge myself that way.

Andrej Zito 

What are you excited about in 2020?

Oksana Tkach 

The collapse of the inevitable collapse of the economy from the Coronavirus? Um, I mean, I guess I had some things that I was excited about a week ago. And now it’s like, I’m excited to be alive for another week. Okay, it’s a joke. Um, what am I excited about? Everything? I don’t know. I feel like it’s going to be great year I have have some really cool new projects. I’m excited to finish them and sort of present them and see those startups that I’m working for. Do while I’m excited about my sort of Twitter translation thing. I would really want to see where that goes. I what are the plans that I had for this year? There was so specific, oh, I am excited to like start really start singing and start sort of posting some stuff on social media. I think that would be really fun. I just joined Twitter in January for the first time. Everybody’s like so tired of Twitter and I’m like, this is cool. I like this. I’m gonna tweet tweet every day. Uhm.

Andrej Zito 

So what is it? What is your Twitter handle?

Oksana Tkach 

Yes, follow me on Twitter. It’s Oksanaissometa. It’s all one word. Yes, so follow me on there. I am sometimes funny. In times of a crisis. Yeah, I feel like in 2020, I will possibly move. I’m planning that not sure how that is going to go. And just kind of probably change my social circles. That would be cool.

Andrej Zito 

Do you mean, you mean outside of Ukraine? Or?

Oksana Tkach 

Yes, outside of Ukraine. I feel like that would be a good step in my career. make a lot of sense.

Andrej Zito 

Where would you like to go? Is it more like US? Like we’re all the relation?

Oksana Tkach 

Yeah, I would like to go to the US. I can’t really explain why something I would like to do. Hopefully, the planes will start flying soon. Before the end of the year. Yeah, I would just like to have that experience and see how it goes. Maybe and maybe it will be like, no, and leave in six months. But who knows? Um,

Andrej Zito 

Do you want to keep your current team in Ukraine and work with them remotely?

Oksana Tkach 

Yes, because it’s really great setup. It’s really outsourcing to Ukraine is an excellent idea. Because we have, you know, a great office setup, the cost of living is much lower and if compared to San Francisco, during the development here is I mean, and we have so many great educated intelligent professionals here. You can you can create a really great team you know, Grammarly basically the whole office in case,

Andrej Zito 

I didn’t know that.

Oksana Tkach 

Yeah, they didn’t, once I really wanted to work there, and they didn’t hire me. Because they do an op, which was probably a good decision on their part at the time. Yeah, so like, Grammarly is a big is a big sort of site. It’s not a startup anymore. It’s a big successful company. But they do basically everything from kids. So yeah, I think I think I would like to be and also I like my guys, I don’t want to leave my company. Yeah.

Andrej Zito 

But then you also said that you want to change your social circle. So I guess you meant people outside of work? Can you maybe expand on that? Like, why you want to change? Like, what are the people that you hang out with?

Oksana Tkach 

I just want to Oh, man, I’m going to offend some people. To Be really careful answering this question. No, no. Yeah, I just want to, you know, growing up in Ukraine, you grew up with with a very specific mindset. And it’s getting a little old for me. And I don’t know, because I’ve never lived anywhere else. So I don’t know how that experience will go. But I feel like I need to have it to maybe feel a bit more understood. Because right now, just my interests and the things that I’m passionate about aren’t really aligning with the people that I hang out with. Yeah. So I’m just I’m kind of like trying to find, you know, my tribe, if that is not too lame.

Andrej Zito 

No, no, it’s not I totally get you. You know, I’m originally from Slovakia.

Oksana Tkach 

Did you feel the same way?

Andrej Zito 

Yes. Always. But it’s also like, part of it is like inside my head. Because I Well, I don’t I don’t think I share this with a lot of people, especially not publicly on the podcast. But like when I was in Czech Republic, because I also call it kind of my home because I went there to study University. But I always felt that like, I need to get out of these post Soviet countries, partially because of the mindset of the people. And also partially because I didn’t feel like I fit in when it comes to how I look like so. Yeah. So yeah, but that was that’s pretty much in my head. You know, like I have very good friends in Czech Republic. So you can always find like, you know, like the niche, think anywhere. But yeah, it’s definitely a great experience to go and live in different countries. I live in so many different countries, lived in Asia and I live in Canada. So yeah, I definitely recommend it to you and also from like, from what you’re sharing with me. I feel like there like a lot of similarities between us. So yeah, you should You should definitely give it a try.

Oksana Tkach 

Which country resonated the most?

Andrej Zito 

Right now I’m very happy where I am. Because of First of all, because of the colleagues that I have. Many of them are my friends right now. Also, because of the work that I do, also, because of the work that I do outside of my work, which is something that I really just want to do, like, for example, like podcasts and creating content and doing and using it as a leverage, like for what I want to do later. And also I just dance a lot more than I used to do before.

Oksana Tkach 

Just randomly dance in your apartment.

Andrej Zito 

Yeah, yeah, it’s so that I don’t have to take classes you know, especially now with Coronavirus. I don’t want to go out. But also the classes here are quite costly if you want to do it, like very rigorously like everyday practice. So that would cost a lot of money. So I just have some like online classes that just paid like 100 bucks for a year. It’s kind of like an online tutorials.

Oksana Tkach 

You can do dance classes online.

Andrej Zito 

Yeah. You have so many things that you can do online. You can you can even do you know, like these cycling classes. There’s like even, yeah, yeah. And cycling. I think, you know, it’s Tim Ferriss, you know, Tim Ferriss, I assume?

Oksana Tkach 

Yeah I have heard of him?

Andrej Zito 

Yeah. So one of his advertisements on his podcast. It’s I think it’s called pelitan something like that.

Oksana Tkach 

Oh, that’s a spin.

Andrej Zito 

Yes. Yeah. So you’re at home, but it feels like I don’t know. I don’t have the practical experience. But it feels like like you’re actually attending a class with other people.

Oksana Tkach 

I thought cycling that you go outside and you’re on your bike, but you are staring at your phone, which seems really dangerous. Yes. Learning to bike online. Now. Yeah, peloton. Yeah, I used to really like spin classes, and you really need that kind of a community because it’s very easy to quit when it’s that intense. So I guess it’s a kind of a cool idea to do that.

Andrej Zito 

Yeah, I think many people, many people need the social aspect to do something. I’m actually quite happy because I can push myself I have the discipline to do things just on my own. Like, for example, dancing, you know, like, I do it every day, like I’ve been doing it for like, the last month. If I do it for maybe six months, in one year, maybe then I can say like, okay, I actually have the discipline. But I think discipline is like very important, because like you can plan everything that you want to do. But if in the end, you just don’t do it, you know? It’s useless. Anyway, things that you change your mind about?

Oksana Tkach 

Yeah, the first one that comes to mind is Yeah, I just I guess I used to be like, I’m going to come and fix everything, and everything is wrong the way it works. I think now, I kind of I have more appreciation for the structures that people put there before, I think it was like, really, it was, I was young. I mean, it was a year ago or a year and a half ago. But when you are in your 20s, that’s a long time. So think ahead, like a kind of a bit of a youthful arrogance. And now I kind of appreciate that, you know, you don’t know what you don’t know, right? That that’s that syndrome. And now it just kind of I’ve come down and I still don’t like when things aren’t. Right. But I don’t need them to be perfect, ou know,

Andrej Zito 

Is it is about is it about not liking or accepting the situation? Or is it more than before that used to blame or judged people for getting into that situation?

Oksana Tkach 

I don’t think I judge them I just didn’t understand was like, why are you doing this by hand if you could automate it, right. Like it was so frustrating to me to see people just make peace with this supplier that they were having. But now I’m like, No, but this actually makes more economic sense. You know, like there are larger forces in work than just the, you know, innovation. That was there was a big one. I have changed my approach to so many things. Because I used to be like, Oh, this is so you know, I have a task and I start to do the task. And I start with, oh, this is wrong. I’m just going to make everything from scratch. And now I go, Well, no, let’s keep everything as it is. But let’s make an addition that makes somebody life a bit easier. That that works for me much better.

Andrej Zito 

Okay? Favorite cat tool?

Oksana Tkach 

MemoQ

Andrej Zito 

Favorite software in general,

Oksana Tkach 

Favorite software.

Andrej Zito 

Or app, whatever.

Oksana Tkach 

What? Okay, let’s see.

Andrej Zito 

Isn’t it Twitter?

Oksana Tkach 

Um, maybe Twitter. Twitter’s pretty good. Pi charm. I’m looking at it right now. It’s like a an ID for programming. It’s really cool.

Andrej Zito 

So what is it? What is the name?

Oksana Tkach 

It’s called pi charm. To make programming easier. What I use is my phone. Telegram. I love telegram.

Andrej Zito  

That’s a messaging I’m right.

Oksana Tkach 

Yes. Because it’s, you know, it’s an nobody saves your data. And there’s basically no limitations. It’s pretty cool. Podasts. I love to listen to podcasts. That is something that I use all day every day YouTube as well.

Andrej Zito 

Favorite word in Ukrainian?

Oksana Tkach 

Um, okay. So when I was, like, 12 years old, I had this weird habit of like, because I was learning English and stuff. I would write down weird new words on my calendar. But I found this Ukrainian word, and it was novela. And nobody uses this word anymore. Or I guess in some regions, they use it. But not verla basically means like, inside out. But it sounds almost like Spanish to me. It just sounded so good. And I wrote it down. And I looked at it like every time I would sit down at my desk at my desk, I would look at it and now it’s just in my memory forever. Yeah. And also it’s it’s represents, like the the Ukrainian phonetic system so well, because we have these open vowels. And these like crawled ours.

Andrej Zito 

Favorite TV show.

Oksana Tkach 

Favorite TV show? Currently, it’s Better Call Saul.

Andrej Zito 

Hmm. Is there a new season out? Or? The fourth one

Oksana Tkach 

Episodes? Yeah, first season. But also, when I was younger, I really used to like Dr. House. Do you remember this show?

Andrej Zito 

I do.

Oksana Tkach 

I was obsessed with it. I watch it.

Andrej Zito 

Oh, yeah. You say when you were younger? Yeah. Yeah. Also, when are you my 20s? Yeah, I was just thinking that I’m just like him. Like, I care about fixing the problems, but I don’t care about the patients that much.

Oksana Tkach 

That’s the last I just recently binged watched it and I was like, he needs some therapy.

Andrej Zito 

Oh, gosh, thank you.

Oksana Tkach 

But okay, so I used to be obsessed with that. And I would watch it over and over and over again. And I would have all of the episodes and I’d have all of the like subtitles and I would read all these complicated medical things. But what I rewatched it recently, I started to finally notice the actual metaphors, you know what it was really about? Because I didn’t I completely ignored them before. But you know, the, his leg pain is like sadness, you know, the key is ignoring his depression. And so the more he ignores it, the more his like, hurt. And then there was this episode when he was in love with Cody, and he, you know, and he, he realized that he was in love, but he wouldn’t tell her. He had this like mosquito in his apartment and in bed him. And he had this big thing that was scratching. And Wilson, his friend was like, do you realize that that is a physical manifestation of love? And he was like, What? Oh, that is so good. Like, every kind of emotion in that TV show is somehow represented through a physical element. You know?

Andrej Zito 

It was so brilliant. Who is your favorite fictional character?

Oksana Tkach 

Martin Eden.

Andrej Zito 

Who is that?

Oksana Tkach 

You’ve never read Martin Eden?

Andrej Zito 

No, I don’t know what it is.

Oksana Tkach 

Okay, so you know, the writer Jack London. He, he like he used to write about like wolves and whatever. Okay. But he wrote this novel called Martin Eden, which is also I just rewrite it again, and my mind was blown. So when I read it, and I was like, 12, or 13. It’s basically about this guy, Martin Eaton who comes his super poor and he comes from this poor family and he, he’s like, strong working on a ship and like very kind of simple, you know, and he meets this woman who is beautiful and rich, and you know, really impressive. And he decides to get smart for her. And so he starts to read all these books and become you know, more and more educated. And then he becomes so educated that he decides that he realizes that like he understands more about politics, and whatever Then other people and he starts to write, and he is really against socialism seems like very kind of libertarian, right? And he is in, he keeps just preaching that, you know, Darwinism, and you know, the strongest should survive and has this belief system. But, and when Oh, and then at the end, this is a spoiler. But at the end, he basically kills himself, because he realizes that he got so smart, that that lady that he was in love with, now she wants him, but he’s like, you’re, you know, you’re a hypocrite, and you’re kind of, you know, not deep enough for me anymore. And then nothing kind of makes him happy anymore. And he kind of goes off. And so when I was small to me, and read it, to me, that was it, like a tragedy, because I felt like, oh, he finally, you know, made a man out of himself. And then he had to die. And that is so horrible. And now while I, when I read it, a year ago, I was like, Oh, this this is like, an oath to socialism, because basically says, He believed in, in, in an individual so much, that he completely neglected that being alone made him so unhappy that he killed himself. You know what I mean? So it’s also very, very, very much like inductor house. It has several layers where you can read it and misunderstand it completely. But when you kind of read between the lines, it’s like it was very kind of a, a troll move. You know, to mask the, to kind of basically say socialism is great. But good of mascot behind the story, but this brilliant man, so yeah, you should read the book. It doesn’t matter that no, you know that he dies at the end it

Andrej Zito 

Is there anything I should have asked you, but I didn’t.

Oksana Tkach 

I think you asked me everything. You know, you know everything about me now.

Andrej Zito 

Okay. I think I think I didn’t ask one thing. When I was stalking your Facebook. I think I I prepared like one question. Do you dance? I think I saw something. Like one video in one video. You were like, moving kinda like this.

Oksana Tkach 

It was like, it was a gift that I was 16. I like made it for my boyfriend. Because I was sick. And we couldn’t see each other. So I made like this goofy dance. That’s probably what I posted. But I do not dance. I am. I’m not physically gifted in any way. I cannot move. Yeah, sorry. I can think though. So you can you can sing and you can dance.

Andrej Zito 

Maybe Maybe next time. Okay, final words. This is your moment where you get to speak to the minds of the localization industry? Oh, what would you tell them?

Oksana Tkach 

Okay, um, start something like start a thing of your own. I think business is kind of miss represented because people that are in business, they tend to tell us well, only the strongest, the smartest, the richest one can start a business. And we also tend to think that the only reason to start a business is to make some money. And I think all that is false. I think there are so many wonderful reasons to start a business apart from making money, anything so many different people apart from like a type a male, white Zuckerberg person that can do it. And they can do it. And they can manage a team or manage a project or whatever, in so many different creative ways. And just because there is no example of somebody like you doing it, that doesn’t mean that you can’t do it. And also think it’s a great way, I started my business because I didn’t like the jobs that I was offered. And I created a job for myself. And I haven’t had a job interview for two years, at least. So I believe it’s like the future of our changing economy is to go and create a job for yourself so that you can do what makes you happy, but also make money out of it. And then eventually, all these small entrepreneurs will kind of connect into this network that works much better than the capitalism that we see right now. So even if it doesn’t bring you a lot of money, even if you even if you have to spend money to do it, but it makes you happy. Go and do it. And don’t ask for permission. Don’t ask for advice, because I mean, just give yourself the permission to go and do it. Yep, that’s my final word.

Andrej Zito 

Thank you. Awesome. This was a real pleasure. Thank you. Thank you for the interview.

Oksana Tkach 

Thank you for having me.

Andrej Zito 

And we’ll be in touch.

Oksana Tkach 

For sure.

Andrej Zito 

Thank you. Bye bye

We’re always creating new localization content

Make sure you don’t miss anything. Be the first to get our upcoming newsletter by joining the mailing list now. 

If you enjoyed that, you’ll love these…

Why hello there!

Enjoy 10% off

on your first course when you join our mailing list.

* All information collected will be used in accordance with our privacy policy. You may unsubscribe at any time.