Natalia 00:10 Good evening, everyone. This is yet another episode of Sunday webinars by Welcome Solutions. In these webinars, we talk with interesting people with interesting careers, often the holders of PhD titles and people who made interesting career decisions and career transitions in their lives. Today, I have the great pleasure to welcome Alican Noyan who will tell us a little bit about his personal story and his transition to the industry.

Dr. Noyan has a background in material science. That’s very interesting because I would like to learn about material science myself and would like to hear a little bit more. Alican went for an industrial PhD program at Hewlett Packard in collaboration with the Institute of photonic sciences in Barcelona in Spain. During his PhD, he developed and fabricated a novel self-cleaning service for Hewlett Packard. After his PhD, he continued working at the Institute developing machine learning models for several photonics research projects. Today, he works for his own company, Ipsumio which uses machine learning to solve problems in physics and healthcare for research groups and companies.

I’m very happy to welcome you, Alican today. And one comment I would like to make is if you guys have any questions or any comments for the content that you saw, then please leave them under the movie. We’ll screen through everything and attempt to answer every question with Alican. Now, I would like to give the floor to our guest so that he can introduce his own story in his own words. Thank you so much, Alican, for accepting our invitation. Please tell us how your story looks like from your perspective.

Alican 02:29 Thanks a lot, Natalia, for inviting me. You explained quite a bit about what I did and what I’m doing. But as you said, let me give more details about my story. I’m based in Eindhoven and I’m living here. I found Ipsumio a year ago together with my business partners. It’s a machine learning company. As you said, I started as a material scientist and switch to machine learning. I will explain a bit about the road. After graduating with my bachelor’s degree, I started my master’s in nanotechnology. At that point, I decided I wanted to do a PhD, but that academia is not for me.

What I was searching for is an industrial PhD physically. And luckily, I found this program in Barcelona by the Catalan Government. They were bringing together companies and universities institutes to solve the r&d problems that companies are facing. In my case, it was a project together with HP in Barcelona, and it was for my PhD Institute. It was related to printers. HP has a very large facility in Barcelona. They have bought inkjet printers and 3D printers. The challenge they faced was the contamination inside the printer. They have lots of sensors inside. And the sensors get contaminated by ink and powders. What we developed was a self-cleaning surface to clean the sensors.

This was pretty exciting for me. But after the PhD, I wanted to switch again. Let’s say I did a little bit of Android development. I decided to try machine learning. I took an online course. And I realized that it’s my calling. I went to my professors after PhD and said, I wanted to apply machine learning to our existing projects in the group. I worked as a research engineer in the group still in Barcelona, and combine machine learning with physics photonics projects. After seeing the value there, I decided to build a company around this idea. This is how it got started. The reason I’m here in Eindhoven is my wife came here to work. I followed her and that’s why I founded the company. This is more or less the summary of what I did.

Natalia 05:59 I would have a list of questions that I came up with while you were talking. It’s very interesting. First of all, you are the first person in the history of this webinar, who did a PhD in industry. That’s already very interesting. I have to say that I know of these PhDs and I know a few people who are doing PhDs in companies, but like in the Netherlands, It’s not very often practiced. I know that many companies like hp and Phillips host a lot of PhDs, and some others as well. But when you look at the percentage of industry PhDs in the whole pool of PhDs, it’s not a high percentage. Before we talk about what the reality of such a PhD looks like and we will switch to business later and definitely will ask you about that, I would like to ask you a few questions about how your PhD looks like?

First of all, I would like to ask you about this decision to go for this type of PhD because you said that you already knew that probably research careers are not for you before you went for a PhD. My first question is, why did you decide to go for a PhD rather than going straight to the industry? Can you explain a little bit about that?

Alican 07:33 The main motivation was: okay, then, if not academia, what should I do? I was looking for other options. What I see was to work on a challenging project and industry. Let’s say r&d, like what ASML is doing, right? What Philips Research is doing so for these projects? They mostly require PhDs. They are also generating new information. It’s not that the industry is behind academia in terms of knowledge generation. In some companies like ASML, for doing research at the edge, they might be doing even more research than academia. This is why I realized that even though I don’t want academia for such a career, I would need a PhD.

Natalia 08:55 If I can say something here, I think it’s really interesting. When you look up the numbers of patents granted, then you can see that the leaders are not the universities anymore. The leaders are companies such as IBM. IBM is the world leader currently, but also Philips and other IT companies or companies producing equipment. I can see that ASML is also very heavy on r&d. The company has the biggest r&d department in the Netherlands. It’s almost 9,000 heads. They have so many PhDs that they even have an annual PhD day. They have a separate fest for PhDs only.

I agree with you and I think innovation in general already is moving to the industry. Unfortunately, academia is falling behind and so I see your point. I understand your point right now. My next question will be more motivation-wise. As you know, once you do a research project in academia, then the knowledge and whatever you produce goes out there as a research paper. You’re the author and you have that feeling of satisfaction. 

But here, once you produce for a company, there is also another factor which is money. If you design self-cleaning surfaces for HP, I’m sure that they probably make quite an amount of income on that, either directly or indirectly. Didn’t you ever have any motivation problems for this reason that, you know, once you work for them, your work is producing a fortune for someone else?

Alican 11:00 That’s a very nice question. This is an important concentration in general, but not for the PhD because it was kind of a decision. You know this and you decide to do it. Because if you are playing the long-term game for three years working on a project for HP, it’s okay. I agree with what you say, if you are working for a company for 30 years, then I would have some motivation problems. But for three years, it’s perfectly fine. Because I’m also learning a lot but not in terms of finances. That is the best option to be at the edge early in my career and learn a lot. That was my main motivation. In terms of that, I didn’t have any problems.

Natalia 12:15 It’s personal. Everyone has an attitude to money. It’s also about your relationship with money and how you perceive money and how you perceive the importance of money in your life, and what type of relations you would accept to have and what types you wouldn’t? I can see that. I would have a bit different motivation structure. But this’s again a very personal choice. My next question would be very, how does such a PD look like? I’m guessing now that since you defend at the research institution, you probably also have some supervisors that work with you from that institution? Am I correct?

Alican 13:09 Let me explain to you how it was. The company comes to you with the problem. They say we have this problem but we don’t know how to solve it. The role of the institute is to come up with a possible approach to this problem. I had my advisor there, Valerio Puroneri. He was my advisor. And we were trying to come up with possible solutions with him. I’m actively working with Valerio. We come up with a solution. I fabricated and tested it in our lab, then I go to hp, and test it inside the printers. This was more or less how it works. I was spending most of my time at Expo in the institute trying to fabricate a soft and clean surface. And once I’m done, I go to the company. We do tests inside printers. There were engineers responsible for me as well.

Natalia 14:31 You have a kind of double supervision. Very good. Did you also produce peer-reviewed papers during your PhD?

Alican 14:43 This was the main disadvantage of this program that the focus is not on peer-review papers. The focus is on solving the problem in a company. They don’t care if it’s a paper or what they just want to get their problem solved. Before PhD, I decided that I didn’t want to go to academia. That was perfectly fine for me. It might be really sad for some other people that they cannot publish. In my case, we had some initial results that looked good. It had to go through a company process for the patent application so that we can publish it. It was not an easy publication. In that sense, it took some time to pass the company procedures and then be able to publish. Since this is a PhD, I have to publish. There was this challenge. But luckily, I was able to publish it and then we also have a patent together.

Natalia 16:15 I guess that there is a little conflict of interest. Because the point of the company is to make as much money as possible. And science is going into this open science mode right now. The point is exactly the opposite of actually revealing all the pipeline and fine-grained detail of the project. I can imagine that it’s hard to marry the two worlds together in this setting.

Alican 16:49 It’s a trade-off and the company has to decide. For some projects, they might say, we’re not publishing anything from this, then you don’t make an industry PhD project out of it. But for some, they might say, Okay, we are interested in solving this but we don’t have the internal capability to solve this, so why not outsource this instead of not solving it? Let’s see what we can do. This is like a trade. In this case, it worked out well.

Natalia 17:34 Let’s move to this period when you are switching between the two careers. I’m very curious since you’re not the first person here on the webinar, who kind of retrained themselves using online courses. There was Marcus Smith here who did the same. He was doing neuroscience. And after his contract expired, it took him three months to take online courses. He did it entirely on himself. He also applied for a job as a data scientist. He is now working in a startup and he is successful. 

Recently, he let me know that he got his first promotion after four or five months of working there. He’s very happy there. His transition was quite successful because he not only got the job but he was also appreciated there. It’s mutual appreciation. In his case, it took him three months. I was curious, in your case, how did you approach this transition? And what was your strategy? How did you choose the courses? How much time did it take you? Could you spill some beans here?

Alican 19:14 The first course I took was the famous course on machine learning. It’s from Andrew Ng. He’s the also founder of Coursera. This was a course on Coursera. He also founded Coursera. This is how I started. While I was working, I was taking it afterward. It took a month or so. Then I decided that I wanted to pursue a career in machine learning. Instead of taking more courses, I went to my advisor to try to apply this to real problems because you cannot learn if you just take courses. This is why I wanted to try to apply it as fast as possible knowing that I will fail. Because there are these things you cannot learn in an online course but you have to start from somewhere. This is how I started.

From that point on, it was always the balance between application, taking courses, and reading books. I also bought quite a lot of machine learning books, statistics, linear algebra, and all these things that help you understand machine learning. There was 66%, application, and 33% learning because I try to apply and see where I fail and then find the related course. At that point, you also understand the course much better. Because you know why it’s important, and what it will solve. And then you directly apply it to your own problem. I’m still learning and applying and I will continue this for some time.

Natalia 21:42 I like the altitude. I think, to some extent, this is exactly the scheme that Dutch universities try to affect. When I came to the Netherlands, I was surprised by how much learning through practice they do here. Because in Poland, when you’re an undergrad student, you get lectures, assignments, and these assignments are usually just closed-form assignments. You have to solve some formulas. It’s something that doesn’t require free thinking about the project. It’s more that there is a convergent task where you have one expected answer. It’s more that you have a little project and there are sometimes a few different ways of approaching the solution.

I became a teaching assistant during my PhD. I was surprised by how much research in some form has already been affected since the first year of study. That’s also why Dutch universities are so good quality universities and they are quite highly rated. I can see why it was so efficient for you to go this way. That’s great. I very much like the strategy. I kind of used the parallel of the strategy when I learned about the job market because I’m learning and then I’m always improving every single time. 

Now I’m learning and I’m running an online course for four weeks. Last week, we had the second session. I was teaching for almost five hours. And during the discussion, I came to some methods that I haven’t heard of, and people were telling me, I should read about this, because like a job market and the strategies to approach job application are almost infinite amount.

There are so many different books and so many different approaches that no one knows everything. But actually, I got a few tips. I started reading and I think for the next edition of the course, I’ll just put a few more slides with even more information because some of these things I found interesting. Some of them may not be as much, but I think like every iteration I know more. It’s a similar approach. I think it’s the best approach. I mean, you will never be able to fully exploit all the things you might learn. 

This is a cleaner way of learning and doing projects is probably the best approach. I know how it happens that you became such a pro in data science. Tell us please a little bit about your decision to start your own company. Was it an easy decision for you? From what I understood, you didn’t have much experience with entrepreneurship before.

Although you were working in a company in Hewlett Packard, so you could see business from behind the curtains in some way. And that was the decision to start your own company a difficult decision for you. What were the factors that you were taking into account? What are other scenarios? Did you maybe consider it? Could you tell us if that was an easy decision or maybe took you a while to come up to the conclusion? I’m very curious about how it happened?

Alican 25:35 I should say that is not an easy decision. But I always had this idea at the back of my mind, like, can I start a company? But, of course, you cannot just start a company for the sake of doing so, you need an idea. I had it in my mind but I was kind of waiting for something. It had to click naturally and you should somehow solve the problem. This is what I was waiting for. It happened when I switched to machine learning and started applying it to my previous projects. There, I saw the value. And I have the chance to experiment with it without founding a company. I stayed off to the PhD with my advisor. The idea was there.

And I was able to practice the idea. I saw that it worked. You still have to convince customers. But then I saw it. This happens a lot. This is the main thing that convinced me that I can do this. Nevertheless, even if everything is perfect, you can always fail. To overcome this idea, my view was: okay, maybe I can give myself a time slot, that I can take this risk and do my best. If I fail, I’ll do something else. This is how I started knowing that I can fail, but I’ll give my best to it. It was not an easy one but a valid thought decision.

Natalia 27:57 Can I ask you, then, what was your wife saying? How did you find your business partner? Because one of the main reasons why startups fail is that the co-founders get into some dispute. It’s really hard to find the right problems so that you can contribute. It’s a real problem that you’re solving. And the second thing is finding the right people to do it with. How did you find the right person? And what was your family’s point of view on your business?

Alican 28:46 First, how did I find Emre? We met in Barcelona through common friends. And our talents were complementary to each other. He is an engineer but he has a lot of experience with management. My life was always technical. And for him, it was mostly the business side, marketing, and sales. This is why I decided to bring the idea to him that this is a team sport. Instead of finding someone like me, I’m looking for someone opposite, so that we can bring our forces together. This is why I decided to open this idea to him, and this is how we started.

As you said, family is very important. It’s not like a decision on my own. We took this together with my wife because this is not an easy task. And sometimes you need to devote long hours to it. And of course, your wife should agree with it. Even if you fail, all these things and all the risks you take together, it’s not like, I can take the risk, and then we shared the risk. We take the decision together. Luckily, she was on board also. This is how we started.

Natalia 30:43 Great. I’m quite happy for you that you found such a supportive partner. It sounds like the beginnings were not that difficult. You found the right person and you were on the same page with your partner. How did you find your first client? And can you also explain to us, since so far, we just know that it’s a machine learning type of project? But could you please tell us more about what your company’s doing today? What type of projects and what type of interest do you have for the future of your company?

Alican 31:24 Sure. We do three things. The main thing is I solved the problem for my advisors and I saw the value. What I offered to him was to find a company that provides machine learning services for his physics photonic space projects. This is what we do is simply in these projects. They build physics photonics devices and measurement systems. To understand all these, you need physical models. And machine learning can help you by building a competing data model and seeing the advantage. Sometimes, you can get better accuracy with machine learning as compared to physics space models.

Sometimes, it’s more resource-efficient in terms of prediction. Sometimes, physical algorithms are very iterative. You need quite a bit of time to get a result. In machine learning, you move this complexity to the training time. But once you train a model, getting a prediction is very fast. These are the two advantages that machine learning can bring to physics and also for other domains. 

This is actually how I got my first client and how we started. But the other things we do are pieces of training. I give machine learning training and also productive elements. These are on these different timescales. Training is a very short-term job. In terms of product development, we work with the hospital to develop an instrument to recognize cells from microscopy images and this is more long-term. This is more or less what I do nowadays.

Natalia 34:10 This is a very interesting model because I can see that you play around with different products on different timescales. And I like that model. When I develop my company, I also basically try to put it on a few different legs so that there is more than just one stream of income and more than just one type of product and there are also different timescales to that. And I make sure that these products are synergistic with each other. I like what you’re doing and I’m curious about all these things. I think the last one is the most exciting. 

After all, it has the actual potential. It might lead to another. It has a huge potential. And I’m curious, like on what stage is this project? And do you think that if you design a certain methodology, then you could also potentially reapply the same way of designing the diagnostic methods for some other problems?

Alican 35:28 Let me explain what the project is. There is a dermatological test called sunk smear test. From skin lesions, you can take cells and view them under a microscope. And depending on what cells, you see, you can reach a diagnosis. The problem was that I teamed up with a doctor from my hometown in Turkey. He is very experienced with this test. People prefer other tests that are easier to use but more expensive. They take more time to use. But it’s easy because you just send a sample and get the result back. In his case, this test was very simple and fast. You need to understand what you see on the microscope. This was a challenge for other clinicians to use it.

You need a lot of experience with these images to be able to understand that. He wrote a book on this to motivate people to use it. He wrote an algorithm. If you see this cell, it’s this disease. If you see that cell, it’s that disease. It was not very easy for people to interpret it because you need to see a lot of images to be able to understand. Machine learning was a perfect fit for this problem. This is why we teamed up and this is how we started. At this point, what we did was a proof of concept model that can recognize these cells. We decided on the six most important cell types and build a model. We tested this on images from new patients.

After we trained the model, we said: this is the final version. We are not going to touch it and tweak it. And then we will test it on new images from new patients not in the dataset. We were able to achieve high accuracy above 95%. This is our current stage. We now would like to add more cells and increase some functionality to it to be able to do this. It’s now like a prototype. There’s a gap between the prototype and the product. We are trying to fill this gap. This is where we are at more or less.

Natalia 38:43 Can I ask a question from a little bit different angle? Because you’re trying to look from different perspectives of the problem. And now I’m just trying to put myself in a scenario where I’m a potential investor and thinking to invest in this project. My first question would be, what’s the potential business model? Because algorithms at least in European Union and the software are very hard to protect. If you develop a laboratory test for testing some cells and there are some physical parts back to it, or some substance or new solution, then you can potentially be patented.

If there’s any hardware back to it and it has to be physical hardware, because if it’s just the software, then it’s really hard to protect by a patent. My question would be, can you spill the beans about how are you planning to commercialize this product and what is the business model or is it a secret? If it’s a secret, please tell us that.

Alican 40:05 I can tell you some things, not all. It’s not easy to protect the software. But in this case, what matters is the data set. Because even if you have the model, you cannot improve it. If you don’t have the data sets, so how we can protect the business. Because it’s never like, we are done, this is finished, there’s no improvement. You will always need to add more cells. You observe some things that need improvement. It will always be an iterative process. You need the knowledge from the professor. You need the dermatology knowledge to exploit this, which he has in the data set.

You need dimension learning knowledge and all these things combined. We think that protects us. There is the question by Rohit. He’s asking if there is a legal privacy problem to get data to train? So for us, we had to apply for an ethical commitment to be able to do this. But maybe there will be additional concerns when we need to turn this into a product.

Natalia 41:49 I think maybe in case you have software, you could also build it in the black box version so that the back end is not visible. Indeed, the model is invisible to the user. They don’t know what parameters you have in the model. I’m now working on the battery of aptitude tests. This is how aptitude tests are made in general. Whenever you go online and fill in one of these personality tests, like 16 personalities, or maybe Gallup strength finder, this popular test, you don’t know what is on the back, and you just get the response on the screen. You cannot reproduce the test. Because it’s pay per click. You pay and you fill it, then you get the response on the screen or send to your email, but you don’t know what is on the backend.

This is a way of protecting this type of intellectual property. I’m just working on doing the same. Doing a public version was back on highly protected. I know that this is probably the only way of protecting this type of information. I have a different type of question now, which is just a very practical question about the corona crisis. Did it in any way affect your business? Because I know that small entrepreneurs have a really hard time these days, at least some of them. Could you please tell us a little bit about did it affect you in any way?

Alican 43:43 For some projects, yes, and for some projects, no. Since I’m based in Eindhoven but working with people from Barcelona from Turkey. All those things were online already before Corona. And they are still online. There was no minimal change for those projects but mostly for training. This was the problem because I was doing them in Eindhoven. They were either online or not happening. But I can always focus on my project. It was not a big problem. But still, there are some missed opportunities for sure.

Natalia 44:44 I can tell because the main source of income for my company was supposed to be on-site training. But unfortunately, you’re doing machine learning. It was an industry. It was not machine learning. I think companies were not hammered that much this year like biotech was hammered much more as a whole industry. I can tell because also I see the statistics of how easy is to find jobs. And I think in your area, it hasn’t much changed in the department. Glad to hear that you didn’t have any extra difficulties. I would like to now hear a little bit more about you as a person.

When you look back at your story when you look back at your career so far, do you have any regrets? Do you think you could do certain things better? If you have, you know, if you could turn back time? Would you do something different? Or do you think that you are perfect in a way of developing yourself, and you’re on the right track? I’m curious what you would answer to this question.

Alican 46:18 This is a tricky question. It’s really easy to say, I did this and that thing wrong. But at the time, that was right. At the time, we were different people. I wouldn’t answer this question, like, when I was 20, I would have done this differently. But I didn’t have this mentality at the time. It’s not correct. To say that, I would have done this like that. No, I mean, it’s not. That’s why I think it’s tricky in that sense. I could have done things better but you need to see all these things to be able to do that thing that way. Without having that experience, how can you do it? I wouldn’t change anything, not because it was perfect, but because I cannot.

Natalia 47:35 It means, you made mistakes, but that was not regrets. Let’s talk about the future. Do you have an image of a perfect life you would like to have in 10 years from now? Do you think you develop into a serial entrepreneur? Or maybe you would, at some point, develop one product that rocks the market in the area that you’re interested in? And then sell a company? Or do you have any picture of where you want to get in 10 years?

Alican 48:20 I have a loose picture. At this age, things change very fast. I would have never imagined, let’s say, 10 years ago that I would be at this point. There is no way I would have imagined this. In that sense, I keep the possibilities open for 10 years later. But there are some general principles I live by. I hope, in 10 years from now, I’ll be still happy with what I do. This is the point. I enjoy what I do especially technically since I’m technically trained. I wish I will do something that gives me joy.

From this technical perspective, I like the thinking that machine learning provides to me for all these projects that I enjoy it. This is what I want. I hope in 10 years and in terms of a bit short-term vision, I would like this sunk smear project to turn into a product and be useful for doctors and patients. This will be a nice thing to see for me.

Natalia 50:06 I wish you all the best. I mean, I like your attitude. This ability to delegate is one thing that may be natural for you, but not for everyone because I think a lot of people have this problem. I liked that you found the person whom you trust to do the marketing for you, and you are willing to share your company with because that’s also not a typical setting like people usually try to delegate marketing to marketing experts. You found this doctor to work with. In this webinar, some time ago, we had a guest, Maria who now works on that analyzes the consistency and the composition of skin products in skincare. She works with her partner. They are co-founders of a company but they are both coming from computational neuroscience.

They’re both heavy on computation and like algorithms and programming, but they didn’t have any knowledge of medicine. She took it on her own shoulders to learn about the skin and the skin composition. For her, the main bottleneck was that the information is very messy and it’s hard to get to hard data. It’s an industry that is very hefty and worth billions of dollars. There are a lot of shady practices she had to learn a lot by herself. I think maybe to some extent, she also slowed down her own project because she didn’t collaborate at that point with anyone.

That was also my question for her that this is a lonely road if you want to get all the expertise by yourself. It’s nice that you found a partner before doing the project. I think this is a very good approach. So what are your plans for the upcoming year then? Because I agree that 10 years is hard to predict. But I think for one year, probably you have some plants on your mind.

Alican 52:50 Before that, let me add something related to this delegation. I mean, with machine learning with these projects, there is no way I can do this without the domain experts. The point is, the way I started was with the physics project, the photonics. I know the domain. I know machine learning and then I worked with experts. For something like dermatology, I had like zero experience. There is no way I can learn it by myself without these domain experts. This delegation was not my choice, but it was necessary. I wanted to add this.

I want to add that maybe machine learning projects fail. Because people do not add this domain expertise to their team. They think that they can solve these problems with machine learning. They think that there is the data and then we apply a model and there’s the solution. This is far from the truth. You always need domain experts. I need them and it’s not my choice.

Natalia 54:22 I see that there are a lot of grounds for machine learning also and I see that in a lot of projects that are proposed, there is someone who has a nice model with nice mathematical properties and they are trying to find as many potential applications as possible. And as long as this sounds good to the grant committee, then they might lend money for it and then there is no validation if this is a viable project. The signing of grants is based on the research history of the author of the grant.

There are a lot of projects but machine learning is very hot right now. Grant agencies are very fond of this type of project. I can see that it’s often the case that the whole career of one person is based on certain models that they are trying to force into as many problems as possible. Even if there is no problem, the problem is made up. It’s really good that you don’t do that. Because that’s not the way to go to design a viable product.

Alican 55:38 I will say that first comes the problem, then the model. The model is not the important part. What’s important is the problem. And maybe the simplest model, the oldest model works best for that case because if you have a great model, very novel, bleeding-edge, it’s useless. It’s useless because there are much simpler models with fewer parameters that can solve the problem. So why use that model? The approach, in my case, would be to focus on the problem and then machine learning comes later.

Natalia 56:28 It’s hard to disagree with anything you’re saying because everything makes perfect sense. I think some people are just not talented for business. That’s what you should do. Rocky is saying, it’s very hard to find the right model. That’s true. But it’s still better to first find a problem and then the model. It’s hard because otherwise, it’s just a flawed business.

Natalia 57:02 Can I add something to his comment? The thing is, to generate value, you have to do many things. The project has hundreds of components. And the model is not only an important part but the most important part. What you need is to find a good model, not the best one, not the perfect one. You need to find a model that works and is suitable for the client and everyone. If you can check these two marks, then you’re done.

You don’t need the last 1% accuracy at the expense of computing powers. Because there are all these other components you have to get right to be able to make all this pipeline work. You need to focus on everything. You cannot just focus on the model. It’s maybe very difficult to find the best model, but not difficult to find a suitable model. And this is what you need.

Natalia 58:37 It’s hard to disagree with anything you’re saying. If you’re solving a problem that no one tapped into before, like, if the classification of these cells was never really automatized, then it’s important to also get the product out and whether you have like 99%, or 99 and a half of accuracy, it’s already very good. It’s already a much larger improvement from what was there before which was nothing pretty much. It’s like in this business intelligence, you have to take the time factor into account.

This is something that we are not being taught in academia, like data loss. You have to make the project and the model as perfect as they can be. They don’t tell us to take this variable of time into account and once you develop a company, you always have to learn the time management and it’s great that you have it. Do you think that this time management is also something you learned in Hewlett Packard?

Alican 59:53 I don’t know. I learned it somehow. If you dissect the problem and look at it, it’s just you see it. It’s not something I learned. What I do is I just take a step backward, and then just look at the problem. And think about it critically, this is the thing, so it’s just done. You realize the time problem. You realize the pipeline problem and all these things. If you look at what you learn with a PhD, then there’s a big problem.

You approach this problem. You learn the approach. And then, you divided it into subproblems. And day by day, you try and fail. No one ever solved it before. Of course, it’s hard. That means you cannot be successful on day one, you will fail and then in the end be successful. You learn to cope with it.

Natalia 1:01:17 Let me now ask you this question about different aspects of running a company. Since you started, is there anything that surprised you about having companies something that you never read in any book about entrepreneurship, or business and something that some little change to your life that you had to introduce, or something surprising that happened to you, or maybe changing your mindset or anything that you couldn’t predict before you started the company?

Alican 1:01:56 I can say that simple things are even hard. I’m talking about non-technical stuff. For technical stuff, I already had the experience but for non-technical stuff, there is nothing easy. You always have to spend time on all the details, even renting a space. You wouldn’t consider it to be a challenge for building a company. It takes time. You have to find the right place. You have to talk with people. You have to see this place. I would say that you have to run the space but you never think about it. This’s just one example.

But then there is all these accounting, marketing, building a website, social media, all these small things that you already somehow know that you have to deal with it but you never really think about it. This was the biggest. It was not a surprise but it was an awakening. This was an interesting thing for me.

Natalia 1:03:29 I agree with you. And you know that the standards are also different. Let’s say if you as a researcher, send an email to some researcher you don’t know, just because you’re interested in the research and you’re interested in collaboration or something like that. When you’re looking for a client for a company, and you send an informational email asking for opportunities and if you have any typos in the email, they would just not take it seriously. They will take it as a scam. Even the smallest detail starts to count. I was also shocked when I realized that I have to be vigilant 24/7 because every little mistake backfires from now on.

It’s indeed much more hustle. I agree with you. I asked this question towards the end of the webinar which is, what do you think you could say to early career researchers who are now at the other point when they consider careers in industry and in general someone who is now in research, and what could you say? Also for people who are running their own companies, is there some particular advice that you could give to those people?

Alican 1:05:08 What I can suggest is to talk with people who you would like to be in their place. If you want to build a company, try to find people who did this, who follow a similar track to yours. If you want to go to a company, let’s say you want to go to a seminar, find someone there and try to speak with them about their experience. This would be the primary. This is what I did. Throughout all these steps, right before PhD, I found PhD students and asked about their experiences. During PhD, I found people working in the company during startups.

Now, people were reaching out to me already who did PhDs but then looking for alternative careers, and they were interested in building a startup. I would say this is very important, not because the things they would say are necessary but you can get a different perspective. You can never get that perspective just by sitting and thinking about it. If you just listen to them for 30 minutes, you can just get something that you never thought about. This would be my primary advice.

Natalia 1:06:56 That’s very good advice. Among all the old methods for looking for jobs, networking has probably the best conversion rates. You can get the most out of it per hour of your time as compared to answering two job offers online and things like that. I agree with you. I mean it’s hard to disagree with anything you’re saying because it’s so rational. I can tell you, sometimes you just meet people who are overly intelligent, but not just intelligent in terms of IQ, so that they can solve some mathematical puzzle quickly, but also have wisdom on different levels. They can zoom in and out on problems.

They can steer themselves on many different levels. They can solve a particular task efficiently. That is very focused logic to solve a task. They can also make decisions on these different levels, and they are good decisions. I’m really happy to see that you’re one of these people. And it’s not very common. It’s really interesting. I’m looking forward to seeing how you will develop in the future and how your company will be developing. And I’m curious but you already explained that you’re not planning too far ahead.

Alican 1:08:38 Thank you very much for your appreciation.

Natalia 1:08:41 Is anyone else willing to ask some questions? If not, then we’ll slowly come to the end of this webinar. If that’s the case, then let me just say, thank you guys so much for attending the webinar. I would like to cordially thank Alican for sharing with us. I think it’s super interesting. And thank you so much for joining us today and for people who are watching this material online. If you have questions, then you can find Alican through LinkedIn. I think this was a great lesson in business intelligence. Thank you so much Alican.

Alican 1:09:51 You can always reach out to me if you have any questions.

Natalia 1:09:56 Fantastic. Thank you so much and Good night, everyone.