Jun 28, 2020 | E008 How To Start a Career in AI as a PhD? Andre Marques Smith on Working for a BCI Startup

Dr. Andre Marques-Smith was born in Portugal to a mixed Scottish and Portuguese background. After undergraduate studies in Psychology, he completed his PhD at the University of Oxford, mapping the development of sensory cortical neural circuits in neonatal rodents. In early post-doctoral work at King’s College London, he studied the role that genes play in instructing neurons to form the right connections. He then shifted to systems neuroscience, completing projects on the development of electrophysiological tools for simultaneously recording hundreds of neurons and on neural circuits underlying instinctive vision.

However, even though Andre developed a career in science, including publishing his work in the most prestigious research journals such as Science, Nature Neuroscience, Neuron & Nature Communications and got all the credits to stay in academia for a lifetime, in 2019, he took a personal decision to pursue a new challenge and direction in his career. After a period of self-study on Deep Neural Networks and Reinforcement Learning, he joined CoMind Technologies to work on building a wearable brain-computer interface.

He is very open about his motivation and life philosophy, and he is eager to help other researchers who are hesitating on how to pursue their careers.

In this webinar, Andre told us how he re-qualified from experimental neuroscientists towards an expert in programming neural networks — all by himself and using only the knowledge acquired online. He also shared why he believes that the startup culture is particularly welcoming to PhDs, and why it is easier to thrive in a startup than in a corporation.

Andre’s contact information: 
LinkedIn profile: https://www.linkedin.com/in/andre-marques-smith/
Twitter profile: @TheFrontalLobe_

Please contact Andre if you need some advice concerning:
(1) Neurotechnology,
(2) Brain-Computer Interfaces (BCI),
(3) Moving to tech from academia.

The episode was recorded on June 28th, 2020. This material represents the speaker’s personal views and not the views of their employer(s).

Natalia 00:10 Hello, everyone. Nice to meet you today. I would like to cordially welcome Andre Marques Smith, our today’s guest. Thank you so much Andre for accepting our invitation. Andre has completed his PhD at the University of Oxford. In his PhD, he was mapping the development of sensory cortical neuronal circuits in rodents. And then after a post-doctoral at King’s College, London, he recently joined technologies to work on using a wearable brain-computer interface.

Thank you so much, Andre, for joining us today. I’m very excited to hear your story from you. Let’s start there. I would be very happy to hear your version of words like, what happened over the past few years? What are the motivations for your decision? And what is your point of view today? Tell us about your daily life as well.

Andre Marques 01:16. Thanks for the invitation. And thanks to everyone for joining. I’m Scottish and Portuguese. I’m Scottish on my dad’s side. I did my undergrad in Portugal, in psychology with a particular focus on clinical Neuropsychology, and some general human behavior as well. It became very clear to me early in my undergrad that I was more interested in the basic science side of things, brain-behavior, relationships, rather than necessarily actually helping people directly.

That changed a lot over the years. But a lot led to me applying to a Ph. D. program in Oxford to be able to do this conversion from psychology into more biological neuroscience. That was a very challenging time in PhD. There was a lot of imposter syndrome. And some issues with experiments are taking a long time to get resolved. That was the first point in my career that I seriously considered leaving academia behind. I didn’t because of some very good mentors that I had in Oxford, because things started looking up and I started getting results.

From there, I moved to King’s College for a postdoc. It was a very productive time. I worked with Rico and Oscar Marine. We didn’t always see eye to eye. But in the end, we got along fine. At that point, I decided I’d done enough from developing circuits. I was getting more and more interested in how the mature brain works and how dynamics in your circus give rise to behavior, perception, and cognition. Around that time, Adam, a lecturer that I’d known for five years and was kind of a big fan of, was moving to the Sensory Welcome Centre in London. It was sort of a hard place to get to both for experimental systems, or science.

It was about sort of enveloping them in the computational neuroscience unit, which is one of the top theoretical neuroscience departments in the world. After my time at King’s College, once again, I was having a few questions about whether I wanted to continue being a new scientist in the long run, or if there was something else that I was interested in doing with my life. I took a compromise route. I thought this is going to be a great place to do experimental neuroscience for sure. But it’s also going to be a very good place for me to pick up some coding skills to improve my data analysis skills to start getting exposed to theoretical and computational approaches.

After a few years there, I decided that academia in long term wasn’t for me. The years that I worked there would leave me equipped very well to move into a different career. At that time, I was sort of thinking of data sciences as an alternative to staying in neuroscience. While doing my postdoc with Adam, I became very inspired to do work in neuroscience, again, started getting involved in more and more projects, and eventually finished the project and applied for Postdoctoral Fellowship, which was a big deal for me because it was the third time that I applied for that fellowship.

And finally, I got it after both initial applications. It’s sort of given me a boost of confidence. I think awards and fellowships are worth what they are. I don’t think, they say that much about your qualities as a person or scientist. They can give you a life raft, you know, something to cling on to when you have those moments of doubt. You don’t know if you’re doing the right thing, or if you’re going anywhere. That’s why that fellowship was so important to me.

This was the end of 2018. I get into 2019 full of speed, energy, and power. I figure out a very productive first six months in 2019. This was working with Hoffer and Scotto and I enjoyed the work in the lab. We’re both very good mentors. I had fantastic colleagues. When I stopped that summer, and when I’m on holiday to see my brother in Norway, I felt that something wasn’t clicking anymore, you know, something wasn’t right. I found myself at some point sitting outside at the midnight, enjoying a beer with my brother and just feeling like I sacrificed a lot. I was sacrificing a lot. I was living far away from people that I love very much, my brother, and my parents.

I had good mentors. I had bad experiences. I saw sides of academia that I didn’t know about. When I first came in, my decision to leave and search for something else wasn’t driven by any particular event. It just felt like every year that I spent, there was one more thing that I saw that just chipped away at this idea of Utopia, which is what I came into the academic career with. At that point, I knew this wasn’t what I was going to do in the long run. I always thought that academic neuroscience was going to be the sort of book of my life. And I realized then in that summer that actually, it was just a Chapter that was finishing, and I was about to start a new one.

But the issue was that the next page in the book was blank. I had no idea what to do after I have seen a couple of questions. I’m gonna start getting into a bit because one thing I want to tell you about before I start doing direct QnA, is the process of me getting out of academic neuroscience and more into machine learning, AI, and eventually, brain-computer interfaces. I think one issue that you can often have with people talking about their career trajectories is that it can be very easy to portray stuff as if there was a master plan. You sort of change things one after the other to get to a particular goal. Maybe it is like that for some people.

I can only speak to myself. It wasn’t like that for me at all. I started just as you would approach a new project in science, when you don’t know anything about it, you start with the hypothesis, right? It doesn’t matter if your hypothesis is particularly good or not. The thing that it does is give you a direction to start doing experiments and collecting observations. That enables you to refine your hypothesis and change your direction a little bit until you nail down on something more accurate. And hypothesis I started with data science.

Years before I have started the machine learning course on Coursera. I didn’t go very far with it because at the time I didn’t have any coding knowledge pretty much. When I came back from that Summer Holiday, I thought I’m going to start, I’m going to do that course, I’m going to take an extra week off, and I’m going to stay home to that course all day every day and see how I like it and see if this is something that I would consider doing. And I did the course. It was very interesting. I recommend anything that I’m viewing.

But there was one topic on the course, particularly, there was a one-shot module on artificial neural networks that just completely caught my imagination. After that module, I felt okay, it’s not data science, I refute the hypothesis of data science. There’s this thing that suddenly I became fascinated by neural networks. Okay, I’m going to move in this direction. That was hypothesis number two. And the action points after this hypothesis was very obvious. This is a very big, complex, and fast topic to which I only had a very small exposure.

I felt okay, I need to know more about this. I need to study more. What I did next was I found some more courses again because he’s such a good instructor. I went into the five-course specialization on deep learning.ai. These are all online Coursera-based degrees. I went and did these five courses between September and December. I was doing this part-time on weekends and in the evenings. They were going well. I was just becoming more and more interested in neural networks. And from the point of view of neuroscientists, that is the sort of intuitions and ideas I could have to offer.

And this five-course series was fascinating, and really good quality. I learned a lot but there were a few things that I realized first that this was the thing I want to know more about. Second that I felt like I was getting a relatively good background from the theoretical point of view, but I wasn’t getting any practical experience. They asked me to put together a convolutional neural network to recognize cats or something like that. I could understand in theory, how I would go about doing that I could not sit down and cut something out.

I just didn’t know how and that got me to realize also that if I wanted to get into this especially to gain this practical experience, doing this 10 hours a week on weekends, or something like that just wasn’t going to cut it anymore. I was going to somehow find more time to dedicate fully to this if I was going to take this study seriously. The final thing that struck me and going to guide my next step was that I knew the sort of general topic that I was interested in moving towards.

I had no idea about what is one thing in the topic that I was generally interested in? What type of roles exists? And within those roles, what are the roles that would be good for you as a person? What are the goals where you can bring something from your background that makes you stand out and you get hired by someone and become good? And at this point, my next point of decision was I’m going to take a study sabbatical. I’m in the fortunate position that I have a fellowship and I’m paid my salary. I’m going to take three to four months out from experiments to work full-time studying and gaining more practical skills.

And this was what I started doing already in mid to late December 2019. I went with an organization called Udacity and they’re not free. They’re a bit pricey. They offer very practical and very hands-on project-based courses in machine learning, data science, artificial neural networks, and a lot of other topics. Their whole thing is precisely helping people who are not in that career to gain practical skills and transition into such a career. They help you out with reviewing your GitHub portfolio, CV, LinkedIn, and that kind of thing.

It was quite helpful. For me here I did a more general artificial intelligence programming with Python nano degree which was a very good instance of learning how to use things like pipe torch and so on. Then I moved on into a deep reinforcement learning course which was also interesting and productive. And here, I was taking care of every project that I was doing on this course to make a nice GitHub page, about Notebooks to treat the whole things as a way to build a portfolio like an artist or a photographer does to show people that they’re able to do the things that they’re talking about.

I would offer one general unsolicited piece of advice which is that if you’re ever in a position where you can use the money to buy freedom, do that. It is absolutely the best use of money that you can make if you’re ever in the luxurious position of being able to. What I started realizing here was that there was this question that I kept asking myself, which was, Okay, I’m getting excited by this artificial neural network stuff and machine learning. But at the end of the day, I’m a 35-year-old neuroscientist with 12 years of research experience in neuroscience. What do I have to offer this field and any employer in this field that is a 25-year-old PhD in machine learning?

What I find interesting about AI and deep networks is something that Andre Carpathia once said about them, but they’re neither less nor more than a new programming framework, item set for writing algorithms, functions and methods, and programming with architectures, activation functions, cost functions, and so on.

This just caught my imagination as a neuroscientist because it made me think, Okay, this is not a programming language. But this is a programming framework of the brain. This is how you build neural programs. That got me thinking about brain-computer interfaces which every neuroscientist knows about, but it wasn’t necessarily something I never thought I’d get into.

I felt like it was an abstract area for me. I started looking into it more. And I started thinking, okay, with the advances in deep learning and data science, I think this is going to be a huge industry. I started thinking about what I wanted to do if I can score a job in a startup, something on deep networks just to gain some experience to learn something about entrepreneurship and starting your own company. And three years from now, I’ll try and launch my startup in brain-machine interfaces. Then I got a contact on LinkedIn from a very interesting young guy called James asking if I was free for coffee. He told me that he was building a startup working on brain-machine interfaces, we had a chat.

I realized that starting a company three to five years from now in brain-machine interfaces was going to be too late. You know, five years from now is when the companies that are starting now and started a couple of years ago are going to be putting products out there. I realized, okay, this good idea. If I do this three years from now, I’m too late to the party. I carried on having these conversations with James. I thought the things you were saying were very interesting. I was a bit skeptical about the technology at first he was planning on using, but I did my background research.

And every time we spoke, and the more references he gave me, the more background research I did on the technologies we were hoping to use, the more convinced I became. Eventually, I decided, I want to be considered for a position with new scientists in your company. When I did that, I realized I want this role. You know, if I don’t get it, I’m going to be gutted. They made me an offer. It was the general software.

It was a very straightforward process, which was something that
immediately caught me on the right foot with the company. It immediately made me realize that I can trust these people. They are direct. They’re straightforward. They’re not trying to rip me off. And the other thing that struck me when I was taking an interview with James and Christina, was the CEO. He was a nice person. Anyone who’s an academic on this chat knows, when you’re giving talks and you’re applying for grants, when you’re giving job talks and so on, it’s very much a top-down.

I’m judging you. I’m questioning if you have these skills, no matter where you are in your career, I suppose, in many aspects in the private sector, interviewing can be that way also, but with James and Christina, it was a conversation between equals. I felt that my skills, my background, my expertise were taken for granted. You know, they will never question in the sense of, let’s make sure you have the qualifications that you say you have. This was another thing that struck me. That made me think that this is going to be a good company to work for because there’s this respect. I’ve been talking for a good while. That’s a sort of continuous story of how I got to where I am now. And we can dissect some things more, perhaps.

Natalia 21:07 Let’s get to the question. The first question is from chat which is I’m interested to hear about how you moved this case. That’s something you covered less.

Andre Marques 21:22 I would just structure my answer a little bit just to make sure I address questions. I went Python from my work with Adam at the ESWC. I did the general machine learning course on Coursera. From there, I went to deep learning.ai for more theoretical knowledge, and in deep learning that AI you cover your typical neural networks, you go into convolutional, you do sequence models, and there’s a couple of modules that are very useful in terms of how you apply these things to real-world problems with advice and how do you know if the issue with your model is that you need more data? Or how do you know if actually, you’ve built the wrong model, your architecture is wrong, or your cost function?

These types of very practical pieces of advice that they offer which now I’m finding useful at the time. This sounds important, but not right now. And then finally, I went to Udacity to gain more practical skills and to start building a GitHub portfolio to show potential employers and some background reading in terms of academic papers. After I did a couple of courses, I was able to understand them. That was pretty much online learning personal projects.

Natalia 23:03 Okay, great. Shall we move to the next question? Can you also throw some light on how students are admitted to the science field?

Andre Marques 23:16 Neuroscience is interesting in that way because I’ve met people in neuroscience from pretty much every single type of background, anything from physicists, biologists, philosophers, artists, and psychologists. I realized that’s not very helpful but I would say generally, it’s a very big field. It depends on exactly what particular subfield of neuroscience you’re interested in working on, whether it’s more human cognitive work with neuroimaging, for example, or if you’re interested in another spectrum, like molecular genetics work.

In the middle, we have cellular work, cells, and circuits connecting. You apply to programs or projects directly. Generally, I think PhD programs are a good idea rather than applying straight to a particular project, which is advertised. Because the programs give you a year where you’re using basic courses, and you just make sure that your background is on the same level as everyone else. And you’re getting research experience. You’re doing short rotations with different labs. This helps you both become a better neuroscientist, a better scientist to learn more techniques and to narrow down what you want to get at in terms of how you are admitted.

I would say research experience is good. Every committee I’ve seen which has been assessing applications, there’s always something they look at. You don’t need to have papers. I think there’s been a trend of people applying with papers already published, which is very unhealthy. It raises the bar of competition to a point where people are very talented in countries where they haven’t had that opportunity and they would fail to hit that bar.

This goes through cycles a bit. I think people are retracting a bit from this sort of level of demand. They’re looking more for people who are smart, who’ve spent a lot of time thinking and reading neuroscience background and identifying important questions. You can do that as an undergrad even without any sort of experimental background. You can do this on your passion and your determination to read and learn. I would say, this is a critical thing that you show in your cover letters and your interviews and discussions with the faculty. If you had an interview and you’re very passionate about this. You know, what you’re talking about, and that you’ve identified a particular field in neuroscience, and a particular question that strikes your imagination.

Natalia 26:26 Okay. I think it’s an interesting topic because I think in neuroscience, there is a big boom. The last 10 years were like so science eating big. There was a lot of money pumped from the public sector into science. That doesn’t serve goods to the students because the number of students went up linearly, but the number of positions didn’t. Now, the competition between new graduates to become faculty members is even more harsh than it used to be 10 years ago. I want to say that there is a lot of programs and it doesn’t mean that every program is a good program.

Sometimes, it’s better to wait and look for longer. I have a blog and I have a post about that lined up for the next Friday. That’s for master students thinking about a PhD. That blog post is the voice of what to look into and what to avoid. The fabric of PhD is not a selling point that much as it used to be 20 years ago. To have a PhD, for the sake of it is not worth it. I always have to think about it because the grass is always greener on the other side. For people who don’t have these fingers, this is something cool to have. When you waited for many years with hard work you have to put it into and it’s a personal sacrifice.

You have to always think about if this is worth it or not. Before you even ask how to get my dream PhD program, maybe first ask, do you need this PhD? And am I willing to sacrifice as much? Do I see myself in science? Or do I think of it necessarily the right way? Because there are many good sides to a PhD.

Andre Marques 29:04 I would add to that, by also asking yourself, what do you want to get out of the PhD? It’s a bit of an unfair thing to ask someone who’s just getting into research. The point of a PhD used to be to find out or should be to find out if this is what you want to do with your life to learn how to become a scientist. But I think because of the factors that Natalia was asking now, it is an important question for you to ask, and to why you want to do the PhD, to become proficient in a particular range of techniques or ideas or approaches that will enable me to get a job in a particular sector of industry, or technology.

For example, do I want to get this PhD because I want to do academic research? These two things will point you to different approaches that you should take different things and you should look for them in a program.

Natalia 30:07 Okay. The next question is how did you manage to move from the theoretical machine learning and deep learning courses to something more practical? How can you be attractive to companies when we only have online courses without experimental background?

Andre Marques 30:26 Udacity does more project-based courses. I’m not trying to advertise them or anything. There are other entities out there that do project-based work. But basically, you’re doing practical things in these courses, and you’re putting them on your GitHub. You’re making nice little notebooks very well commented with all the cells running correctly identifying what you’re doing and theorizing a bit about why something is working in a particular way.

A lot of people make medium posts based on these projects. It’s a good way to summarize yourself what you’ve learned about something and sort of start building a little online presence in terms of your expertise. Another thing that was very helpful here was something fairly strategic that I did, which was doing this for a reinforcement environment. Because reinforcement learning has this cool thing where open AI set up opening gym environments which are video game environments, where you train agents, using different techniques in reinforcement learning and AI to beat the computer to a range of games, anything from space Invaders to whatever.

The cool thing about Reinforcement learning is that you’re working in these simulation environments. It’s not necessarily the case that you need to have a huge computer with a 10,000 pound GPU, and gigantic data sets, or take days to train your network. You can do some pretty cool stuff. You can experiment with a lot of really neat little things in deep reinforcement learning with a decent average gaming computer, or even a laptop and go into AWS, or use Google TPU. I think that’s the first part of my answer. Building up a practical portfolio, a nicely maintained GitHub page is important.

One of the things they want to see in companies is that you can work with other people because you’ll always be developing stuff as a team. The time you spent commenting code, documenting your functions, writing docstrings, all these kinds of things, is worth it. They look at these things, they want to see this more than a piece of brilliance in how you solve a particular problem. It doesn’t matter to them if no one’s going to be able to work with you. Another thing I was gonna say here is, it depends on the companies. This was my experience and also my perspective.

I think startups can be very good, specifically for people who are coming from a different background, such as experimentalists and trying to get into something more like machine learning and deep learning because big companies like Google, Facebook, Microsoft, and Apple have this luxury, where there are hundreds and thousands of people knocking on their door and applying every day. On every other job posting, they have the luxury to filter completely people who only fill out 100% of the requirements, and then some on top of that. They get to be able to look only for people who are already the finished product regardless of potential.

This infuriating is very frustrating to anyone because there’s so much talent out there. I would say an interesting alternative is to look for startups in the field through LinkedIn, do a lot of background research on Google to see what interesting little companies are out there because startups don’t have that luxury. They can pay well but they’re not going to pay the same as a big tech company. They don’t have the same name and reputation and not necessarily the same stability. It means that they don’t have the luxury of looking only for the final products. They will take you on board even if you meet half the requirements or 60% of the requirements or whatever, they accept. You are going to grow with them.

The other thing about the startup is that in a small company, there are other qualities, if you don’t have the technical background, there’ll be other things you can offer. You’re a good team player, or you’re very versatile. You can wear a lot of hats. The case of an experimental background has been extremely useful. It’s a coal mine. I’ve been a bit of a journeyman in neuroscience, like, I’ve done a little bit of genetics, I’ve done a little bit human work, quite a bit of cellular and circuit mapping work. I’ve done some systems work.

I wouldn’t call myself a specialist in any of these things because I’ve picked up a lot of different items here and there. This means that there’s a lot of stuff that is now become quite useful, stuff that I know or stuff that I’m aware of. This background can be useful. It’s not necessarily a drawback.

Natalia 36:23 I think it’s also worth mentioning. The fee by itself doesn’t really give you much. It just depends on what you do with it. It’s more about navigation skills and how you’re able to combine your previous experience, use your strengths in a good way, and find an argument where you can use your strengths. Avoid environments where you know that your weaknesses will come out. It’s still about every job. You have to first know yourself well, and what you’re good at and try to fit into a place where you can use specifically those skills. I have a question?

I myself had that very similar dilemma four years ago. My contract got expired before I decided to do what I’m doing right now. I also know, I had the training that was very technical because I have a background in mathematics and physics, and I was doing also computational neuroscience. I was programming during my PhD. And my CV looked like I will be just another machine learning engineer or data scientist, and wherever I was going, I was also automatically classified as such. Even though I am a person, no one looked at that and gave proposals over a data scientist, and I was thinking, this was kind of an easy way out for me because I had the credentials.

I knew that I would be paid well from day one. And I was kind of, you know, welcome most with open arms to what I had in my CV but I might be concerned that I will start if I do this type of job by becoming more fungible in a way. I have to program something. It doesn’t matter for them who does this as long as it’s doing well. And it’s done well. It’s either that you solve the problem and your solution works or this type of job doesn’t work out in your style. Put your personal touch to it and notice the main problem. I thought that It was hard to predict how the job market will be working. I have a gut feeling that in the long run, people have the professions today where they can put a personal passion into what they’re doing and they can put their name on it, like a graphic designer.

This is like the same as you know most programmer or graphic designer is working on the computer. But a good graphic designer who has a unique style can leverage on the personal style and then be recognized for the style. They’re respected for what they achieve. A programmer is always busy studying programming. How do you feel about your second non-fungible type job? Fungibility means that you’re replaceable.

Andre Marques 40:16 In which job?

Natalia 40:23 I think that many jobs are related to coding. The types of technical problems where you have to make a project of a solution.

Andre Marques 40:37 I think it depends a lot on the context. I think, in a small company, like a startup, it depends on which person particularly you bring in because everyone on the team needs to contribute to the common good. You do get a lot more of the person’s individuality in their positive and the negative sides shining through. I think working in tech and deep machine learning, there’s a surprising amount of creativity that is required for these positions. There’s a surprising amount of analytical thinking that you get from your background as a scientist and experimentalist even potentially.

There’s also a sudden flash that you need to have to be able to look at the problem differently and solve it. And this is something where I think, someone coming from a completely different background often has an edge.

I mean, I’m sure I look at deep networks in a way that is a bit different. I’m always drawing the comparison between neurons and circuits. They’ve changed the way I think about them. I’ve based my thinking on these things in terms of my previous thinking of brain cells, and so on. I think it’s better. I know what you mean. I agree with what you’re saying that it can’t be like that. I think it’s better in the ramification of the sector that you’re looking at, in the context of the company. And I would add something related to what you’re saying, which is the thing where we can be attractive, which is that we tend to think of soft skills and hard skills.

Hard skills are called hard skills because they’re really hard to get. You know building deep neural networks, linear algebra, this kind of stuff. And soft skills are these fluffy things that don’t matter. It’s easy. It’s the other way around. Because if you need to learn linear algebra, or you can read a book and you do exercise. If you want to learn about networks, do the courses that I’ve told you about. You want to learn how to manage scientific projects on the timescale of three or four years, guess what it takes three or four years to learn that. It takes doing that to learn that if you want to learn how to be able to absorb a lot of financial literature very quickly and summarize a complex field in an abstract, you can learn that and it takes about five to 10 years. There’s no book about it.

A lot of these things that we pick up tacitly come from academia and particularly experimental backgrounds. These are things that end up being surprisingly useful going forward. I think it’s important in making yourself attractive to companies in these domains to reflect yourself. Be fair with yourself in recognizing how important these skills are and which ones you have and explaining them in plain English in your LinkedIn, in your resume. Because the chances are, you’re so used to thinking in certain ways and having certain skills that you don’t even see them as skills anymore, but they’re incredibly valuable. Some of these skills are the skills that I’m using heavily right now in my new position.

Natalia 44:16 Great. Shall we move to the next question? We have quite a bunch of questions today. Someone is asking, I would love to hear a bit more about your work and specifically your role there.

Andre Marques 44:33 We’re working on wearable brain-machine interfaces. We are both a hardware and a software company. We’ll be building prototypes to image the human brain non-invasively. We think we have a particularly good technological approach to do this at a high spatial and temporal resolution and non-invasively. This sounds like the Holy Grail. And I can’t go into details for obvious reasons. But basically, my role is some sort of sitting right in the center of that company.

On one hand, I’m collaborating with our hardware engineers and electronics engineers who are building the physical prototypes. They’re the guys who know about integrated circuits and optics and all these kinds of things. They don’t necessarily know about the brain. They’re working with me, and I’m telling them, okay, to decode a particular type of motor activity in a certain task, we will very likely need this level of spatial and temporal resolution, we’re going to record from the different brain areas. You can start to see how these requirements shaped the hardware development that has to happen, and you can’t just tell an engineer building something to image the brain. You need to fill out a specific sheet.

That’s one of my job requirements working with them in developing this hardware. The other side of my job is decoding work. On the other side of work, I’m facing the machine learning and AI side of the company. And here, we’re looking at analyzing the signals that are coming from the prototypes from the devices and extracting their meaning. We’ll have some demo tasks we want to do, for example, we may want to classify the presence of a certain visual stimulus based on patterns of visual cortex activity. There, again, I’m working with machine learning and AI engineers who are not neuroscience specialists but they know this stuff in deep neural networks and other aspects.

I’m not a specialist like they are on machine learning but I can speak their language. I know neuroscience and I can interpret what we need to do, what we need to decode, where we need to do it from, and the characteristics of the signals. These are the neural point of view. I can talk to them in terms of structuring our experiments, structuring our data collection or data structures, what types of decoding algorithms, and what type of neural network architectures might be most productive for doing this.

Through that dialogue between me and scientists, who know a couple of things about deep neural networks and can speak the language, and the expertise of a machine learning engineer who knows a lot about these things, the magic starts to happen. I would say any scientist was potentially interested in this area of machine learning neural networks. Use your own domain-specific knowledge as leverage. This is important. Can you become an expert machine learning engineer at the same level as someone who did a PhD in that? I’m not gonna sit here and tell you what you can do. You totally can, if you put your mind to it, and you invest the time.

When you start applying AI to scientific data, to things like neural activity, decoding and collecting that data is not just expensive. It’s really hard. Designing the experiments, collecting the data correctly, structuring the experiments so that your data and your model can generalize and be robust is very difficult. This requires a huge amount of expert knowledge about your domain of application. And this is true for a neuroscientist working on brain-machine interfaces, as it will be for chemists working on drug development.

I’m literally in the middle of the whole company. It’s very nice because I get to see and understand what is going on on all sides. We all do because we’re still a small company. And that’s a great thing that you feel, how your work connects to everything. But particularly for me, I’ve been learning a lot about hardware developments and optics and signal processing, for example, frequency modulations, and all these things that engineers do, which are amazing. I’ve been sort of using my neuroscience background in picking up on the learnings that I’ve had from deep learning and machine learning and I’m also learning from my colleagues who are experts in that as well and furthering my expertise in that direction.

Natalia 50:02 Great. I think in the process, we also asked the question about how to combine experiments.

Andre Marques 50:14 I would have something very quickly to that, which I think is, on the academic side near scientists, this division is going to disappear. I think as an experimentalist, we’ve kind of been playing as scientists in neuroscience. You know that you need better. You need more sophisticated, more quantitative theories for fields to really push forward, not just the sort of naive by biological conceptual theories that we have. And people are getting trained this way. Nicole and her generation are starting to be trained in these two aspects. Finding the right balance is going to be a personal choice based on your interest but you’re always going to have a bit of both in a proportion that is based on your passion.

Natalia 51:09 I can add something to this, what I see in using your science papers, I think that most of these groups and impactful groundbreaking publications were put this article kind of thing crumbled. Experimentally, squirrels are very good statistics, optimizing their own data, rather than just relying on someone else to analyze it.

Andre Marques 51:38 It also shapes your thinking. It makes you think differently. I’m reading right now The Book of Why by Judea Pearl and Dana Mackenzie. While I’m reading that book, I keep thinking that I wish this had come out when I was 20, or 21. Because it would have changed the way that I did science. I won’t say more about the book. Google it and devour it. If you’re a scientist, it’s a beauty.

Natalia 52:08 Great. Shall we go through this question? How did you deal with the feeling that started with less knowledge than other people who studied directly?

Andre Marques 52:22 That’s a great question. And I sort of went into it a bit already. But I’ll summarize that. What I focused on was how could I leverage the 12 years that I spent doing experimental neuroscience in many different domains to put me at an advantage over those people. I leverage the domain knowledge. I couldn’t beat them by knowing more about machine learning and AI than they did. It would take me years to get that. Eventually, I would, but it wouldn’t be what I would want to do.

What I tried to become was someone who spoke both languages. I think when you’re dealing with career transitions, that’s critical to build something about you that is unique and that makes you stand out from the competition. And, In my case, I think that was the thing I highlight. I can speak this language, I can build models myself. I can code these things but there’s not going to be produced in another way.

I can build things. I can make experiments. I can test models and I can build stuff that works. I can communicate with these people. I can break down the neuroscience problem into a machine learning problem. And then I can collaborate with the machine learning engineer to achieve something that neither he nor her alone could do.

Natalia 53:53 Okay, great. The next question from Miguel is, can you work on your own projects ideas? And also how do you stay up to date with AI advancements?

Andre Marques 54:10 We’re a startup. We’re a small company. We have a very well-identified mission that everyone on the team is incredibly passionate about. But it’s a very difficult thing to do. That’s our goal. We come up with our projects, with our ideas, with our own hardware or software to tackle this. We have brainstorming meetings and we propose research projects and discuss them. These things come from individual people and you can work on something that you need to be contributing to the overall mission. Everyone has things that they need to do at a certain time. But you can spend time on projects that you build up to help you achieve your ultimate goal.

Saving state on AI advancements is really hard. It’s so fast. It’s even worse than neuroscience. It’s just the speed at which stuff comes out. I’m already starting to feel outdated on a few things. I was studying these things, at least in reinforcement learning full time up until April and some stuff came out on unsupervised contrastive reinforcement learning that I feel I should be reading about.

And I have like, 15 browser tabs of stuff that I want to read. Sadly, I don’t have a good piece of advice there.

Natalia 55:38 Maybe, this will be also a business idea.

Andre Marques 55:47 There’s a thing which I find super useful, a newsletter. DPI sends you a digest of exciting papers and a little summary every week. This is very useful. I would recommend that.

Natalia 56:07 I’ve always had a lot of new startup numbers. After two or three months, they always start talking. When you asked, what you do, they start saying, we do that, and our product is our mission.

Andre Marques 56:37 I would add that when you have a small team, and you build cohesion, this is what starts happening, people stop talking about I and me, and they start talking about we, and this is well spotted. It’s a wonderful thing. It’s so different from how I was used to working by myself as an individual and maximizing my research output. We’re all feeling part of a team and crunching in other stuff problems. It’s great.

Natalia 57:13 It would be ideal if this talk was like this. But it’s not the case. You’ve got some factors for success. We think that there’s a specific thing about you or maybe, about your purpose, or anything else that you might point out as a potential reason why this is. I think the founders like CEO and COO are two of the nicest people I’ve ever met. I haven’t heard my CEO say a bad thing about anyone. And it’s genuine. He’s genuine and a nice person.

Nice people tend to seek people who have similar values. I’m not saying I’m very nice. I’m not. But they tend to be good at building environments that you want to work in which bring out your better qualities. It’s all about people.

Natalia 58:19 I think it’s two things. The way I see it is probably a combination of personality, but also comparable because of also a lot of toppers. If you somehow managed to get your first investment, get through this first. Sometimes not a very competent person makes it through and then is also able to create this startup culture. It’s great to sit with them and talk about ideas.

When it comes to execution, they do not necessarily know, they elevate the work to you because they don’t have the days to do everything. They will be able to do it too. It’s just that they have to delegate but not clueless people. They have to be specialists.

Andre Marques 59:17 I agree. I think if they write their ability to how they structure the work, how they manage, how they do the project management, and how they plan, then they can be the nicest person in the world. But they can end up putting you in a situation that is impossible and will destroy you. You’re right there.

Natalia 59:39 The next question is do they have any vacancies because it might be after some of the participants are already checking out LinkedIn.

Andre Marques 59:55 We have some vacancies at the moment.
I think we’re still looking for a Signal Processing Engineer. I think we’re still looking for an optical engineer. Keep a check on our website. We’ve been keeping the job posting more or less up to date. There’ll be more space for near scientists probably later in the year. But I don’t know right now.

Natalia 1:00:23 Okay. Are there any recommendations for choosing a good mentor?

Andre Marques 1:00:34 Great question. People tend to go for the really big names and the senior people. Consider small labs, they will have much more invested in you as an individual. If you’re in a lab of 50 people, you’re not going to be doing your PhD with the lab head and going to be with some postdoc, which may be good or not. If you go to a small lab, they have an investment in you. Everyone needs to succeed in a small lab. Another thing I would say is, to try and see if they are still doing experiments, or if they’re not, do they seem sad about that because that’s a really good indication.

Those are the people who are gonna be very passionate. They’re gonna help you a lot when you have technical difficulties. They’re going to come in and be like, okay, I’d rather fix this problem and be marking exams over there. Let’s do it. And talk to not just the current people in the lab, talk to people who used to be in the lab. They don’t have anything to gain from saying good or bad things. But from a human point of view, people will usually want to talk if someone’s been very good to them. They want to say good things about that person, and they want to repay that.

Someone’s been very bad to them and they want to prevent someone from making that mistake. Don’t underestimate it. Talking to current people can be difficult as there’s a big conflict of interest there. But look for people who did their degrees in that lab and approach them. And they will talk to you.

Natalia 1:02:28 I could add to this, if you want to know about the dirty person, then you can always ask something like, can you tell me three good things about the person, then they will probably tell you what they didn’t like because they will kind of excused themselves. They also say some good things so that they will feel less bad about this. If you ask this way, they will be more honest.

Andre Marques 1:02:58 I think that’s a really good trick.

Natalia 1:03:03 Okay, we have a lot of questions. Next question is, how did the actual practical match the expectations you’ve thought of it based on the theoretical knowledge gathered from the courses?

Andre Marques 1:03:19 I would say so. I think that’s a good thing about viewing courses. There’s a Stanford professor who works with a lot of companies. He has a lot of commercial and industrial expertise and experience. He brings up practical pieces of advice through his courses which are things that I find myself now thinking, that’s what he meant, and this is what he was talking about.

They are more theoretical and less project-based. Most courses are on programming and AI. You’re learning how to make something.
You’re learning how to build something. They’re already that theoretical kind of practical.

Natalia 1:04:14 Okay. The next question is what is the better place, to begin with as compared to the bigger established companies?

Andre Marques 1:04:30 I can only speak from my experience. What I felt very deeply was that this was the right thing. Because, you know, one of the key reasons I came into academia was I didn’t want to be just a cog in the machine. I wanted to do an individual contribution to something no matter how small it was. It was recognizable. It was me and only me who could have done that. Startups have this element because I feel if it was a guy in this role, they would be doing different things. They would go in different ways.

The company would go differently. In a bigger company, there are a lot of established cultures. There are a lot of processes. There are a lot of very big companies that end up having a lot of things in common with academia. But there are things I feel very similar. And I’ve got friends who work for big companies and there’s a certain type of politics and organizational behaviors that starts happening when a company grows above a certain range.

We did not evolve to deal with networks above a certain number of people. I would say, from a biased point of view, I think it’s easier to get the foot in the door, in a startup. And then you know, you may or may not fit well in the particular logic of a startup. But as getting the foot in the door goes, it’s something that can at least enable you to gain experience. The startup for that person can be a punch pad.

Natalia 1:06:27 If I could add something to this, it’s also a cultural problem. There are different tribes, and the rules are different. I also make people feel better in cooperation because cooperation every day operates a bit like legal research field, then you have to work on your own CV, your own portfolio, and you have to take care of your own promotion. You have to kind of apply for positions and apply for tech courses. It’s like a very similar career in cooperation, as there are many ways of working. And the difference is that you have a contract. In that sense, it’s safer. But in other ways, you are still like exposed to high competition in less teamwork and more self-promotion than startups.

Startups also have downsides. For instance, it’s a small place where usually you kind of text your boss, direct boss, and if you don’t connect with your buyer in Boston, there is no way around and you cannot sit somewhere else because the company is small. This is a cultural problem. It’s also up to you to discover what type of rules fits you and your susceptibility to stress better.

Andre Marques 1:08:07 I think you nailed it.

Natalia 1:08:12 Okay. Is it possible to get into PhD without any publications in formal fields you were in?

Andre Marques 1:08:22 I didn’t have any. It was a while back. I revealed my hands already. It’s fine. I got into my PhD in 2008. It goes through cycles. I think there was a peak. I’m talking mainly about the UK system. But there was a peak maybe five years ago or so where suddenly there were undergrads who had a second author. I saw some of these CVS, and I thought this is ridiculous. This is an anomaly. It’s not fair on anyone. It’s a product of the fact that you are fortunate enough to be born in the United States and go to Harvard for your undergrad and do a project rotation in the lab. If you’re coming from India, or Spain or Portugal, or a country, that doesn’t have the same economical investment in science. Some countries are fortunate to have the same thing.

You need to have very good letters of recommendation. You need to show in your CV a real commitment that you’ve tried to involve yourself in research projects to get research experience throughout your undergrad. You’ve spent a lot of time thinking about these topics. You need to write a really good cover letter and a very smart cover letter. You bring an original spin to it. That’s the other thing. Really impressive people come in with a different way of thinking. You’re not simply regurgitating a review paper.

The thing is that the academic world is a bit more varied in terms of locations. People we all have in our heads institutions like MIT and Stanford, Columbia tend to think so. There’s a wide range of really good universities in countries that some people don’t even necessarily think that much about. Switzerland is a fantastic place for science. Spain has a couple of very strong Institutes. Portugal is also a great place. There are also a few great places in France. It’s really about the people who are doing really good work in that area? And yes, they might be one of the elite institutions.

They might be an institution that people don’t even necessarily think very much about. But if you go and do your PhD there, you will do amazing things. The competition level to get in terms of competing against people who have papers already might be lower. And you might be getting into a better situation.

Natalia 1:11:11 Great. The next question is from Maria. Hi, Andre, I’m a neuropsychologist in Delhi. At some point, I want to start my own startup in neuropsychological rehabilitation and therapy. Do you think having a PhD in neuroscience will be helpful in the next logical step?

Andre Marques 1:11:27 I think if you do a clinical focused PhD, and particularly focused on neuromodulation techniques like TMS, transcranial electrical stimulation with alternating currents. Ultrasound is becoming starting to become a really big deal. If you do a PhD that is targeted and is teaching you skills, which are going to be useful for your startup in your psych rehabilitation, then go into it very focused and come up with ideas for a project that you’d like to do. That is useful to you as a person and as an entrepreneur when you start. I think it would be helpful.

Natalia 1:12:18 Right. Do you have any recommendations for courses or specific projects for this year?

Andre Marques 1:12:28 Look at open BCI. They’ve got some datasets. They used to have competitions. They’re a very interesting organization. But they have data sets on which they have challenges to see whether you could decode finger tapping movements. And that’s an annoying thing about BCI. There isn’t like there’s maybe one or two textbooks, which are like PCI 101. There’s one called introduction to neural engineering. There are really good resources from 2013 already. It starts to show a little. There’s another by Wolpaw and it has been more in the team. They also have a really good textbook on BCI. I was saying neuro programming is going to be a topic.

Natalia 1:13:46 I would like to say that I support all the endeavors to start the company. As a researcher, it’s cool to me. I can tell you and that is fine as well. I made a resolution of that. Anyways, the next question is from Alexandria. Andhra, would you consider switching to a job in academia working on BCI engineering? Or do you think the startup environment is better aligned with your goals or interest?

Andre Marques 1:14:32, if I had to pick one thing, that was the final thing that made me flip and decide. I started realizing that I’ve been publishing pretty well and neuroscience is the project I was doing at the time. It was going in a really good direction. But I started looking at that project. And that project was like a sub example of a phenomenon, which I was generalizing to a larger process in cognition and neural information processing happening in the brain.

Starting from the top, and then going through a very specific implementation that’s how you do if you’re working as an individual. That’s how you have to do it. If you’re working 16 hours a day, there’s only so much you can do as an individual. If you have five or six people in a team, don’t even take more than that. They’re working on the same project and their interests are aligned. They’re working equally. Then the type of stuff you can tackle is completely different. You take on the big problems. I think, working on BCI neuro-engineering in academia is fascinating.

Some great labs are doing fundamental work. But each person has their baby project. They collaborate on some stuff. Collaboration doesn’t mean I’ll do some stuff for you and you do some stuff for me. That’s not what the word means. The startup environment isn’t like that. You need people to align and become bigger than the sum of the parts. And to me, it felt like I want to spend the next 30 or 40 years of my life working. I cannot tackle that size of the question. In an academic institution, that became very clear. I could only talk a little bit about that.

Natalia 1:16:53 Okay, great. The next question is do you think that what you gained by switching from academia to industry something is generalizable to other fields, including other new science fields within psychology?

Andre marques 1:17:12 I think so. I feel I’ve gained that there is a difference that I learned a lot from my previous colleagues in my previous labs. But they were still your scientists. Now I’m working with optical engineers, signal processing electronics engineers, and I’m having to read things that I am completely unfamiliar with. I’m picking up all different kinds of knowledge and skills. If I feel I’m picking up knowledge at the same pace, I was picking it up in my PhD in the beginning. And I think that is part of what you get potentially from switching from academia to certain fields of industry.

I don’t think it’s not a neuroscience-specific thing. There’s a lot of psychology that they do right now with all social media algorithms and all these attempts that manipulate our behavior through clicks. People who are using that knowledge for the good are not bad. There’s also very nice work going on in these areas. I don’t see why it shouldn’t generalize.

Natalia 1:18:31 I have a question. I see that people for the first few months are sweeteners. They are like on drugs. They’re like jumping on the couch and singing under the shower. Do you think that this is enthusiasm or just this honeymoon effect? Or do you think this is because you found your way of living?

Andre Marques 1:19:12 I think it will last. I think it’s just because I found somewhere where I have a huge degree of personal compatibility with the people I’m working which I’ve had before, especially my mentor, Simon, but for my PhD, it feels a bit like that time and personal compatibility. And I’ve also found somewhere h the niche for my role. I found the intersection of artificial intelligence and neuroscience. It turns out the thing I dreamed about, and it’s only going to exist more and more now. It’s becoming a bigger thing.

There’ll be harder times in the relationship because there are ups and downs. But as long as you both want to come back and meet in the middle, go forward. I think that will be the case unless we screw up very badly. I don’t see these critical things that make me happy at the company. I don’t see them changing because they’re their core aspects.

Natalia 1:20:26 I recently wrote a blog post about my honeymoon with my new job. I just put the link in the chat. I have the same state of mind. That is a bit like a grant of two months. I’m quite positive about it. We have two more questions. And that will be wrapping up because we are already over time. But it’s an interesting conversation. The question is, now that you’re working on the company, do you see yourself starting your own company at some point? Do you ever consider joining startup schools programs like entrepreneur first?

Andre Marques 1:21:20 Our CEO didn’t. I’ve heard good things about it. I’m 100% focused, committed and dedicated to what we’re working on right now. I like the idea of, starting something from scratch. There’s a certain romanticism to the rich, which appeals to me a lot. There are challenges associated with that. I don’t know if it’s something that someone in their 40s might have necessarily the energy at that point for but it sounds cool. I think one day, I would like to find out how that feels.

Natalia 1:22:17 I’d also research about how was the success rate of the company, given the age of the founders. People are more reasonable, experienced, and better connected. I feel more successful. It maybe makes such a thing that makes complete sense.

Andre Marques 1:22:53 I was gonna say something about this. I forgot.

Natalia 1:22:59 I also see, once you have that thought, and you try to put yourself in the shoes mentally over entrepreneur. This is like a bug in your brain that will never leave your head until you die. It might kind of eat you for years and years. But eventually, you want to try that. I struggle with it for 10 years or 20. But you sooner or later will try.

Andre Marques 1:23:27 I remember I was getting common which is one of the good things about working in the company. It demystifies something. I used to think that to start your own business, be it a startup or not, you got to have the idea. You have to know everything about the idea, about implementing it, and everything that the company is going to do. I’ve been realizing that’s completely false. You don’t need to have the idea. You need to stand it conceptually and know what you’re getting into, but then you need to hire the people who complement your weaknesses to get it done.

Then you go and hire the electronics engineers and the neuroscientist and the data scientists and so on. We know how to do specific things, but realizing this and seeing the way our company and our CEO work, it’s been very empowering. You just get this feeling that coming from academia, there’s a lot of gatekeeping. There’s a lot of staying in your lane. You don’t know about this and you don’t know about that. This is our field. You don’t come in here. Who are you to come up with an idea? How dare you write an original thought? And the culture where we work and many other places is just completely different.

Natalia 1:24:54 It’s also students in research. We are supposed to professionalize our projects before they get published. They have to be really of top quality and packable to be published. But it sort of doesn’t work like this, you just make the most rudimentary working product, and put it there and put a price tag and then play around with the price tag so that before she is willing to buy it, and then just listen to the user’s book, you try to get the user as soon as you can, and put the price. It’s just different. It’s just like jumping into the deep water.

Andre Marques 1:25:38 I think sometimes, people misinterpret this as not being rigorous enough, or something like that. It’s not about the two different things in academia, where every single detail could potentially matter. You don’t know, it could. You spend a huge amount of energy and time going into every single particular detail, that could conceivably matter. That means you end up losing a lot of time on stuff that ultimately in the end you didn’t have to get far in startups. I don’t know about the corporate result, but you need to be focused at every point in time, you need to be asking yourself, okay, what am I trying to achieve with this? What is my focus? Why am I trying to get out of this task? and you’re separating the wheat from the chaff at every point because you have limited time to get somewhere.

You need to cut corners, but the wisdom comes from knowing which corners you can afford to cut, and which are just going to lead to loss of rigor. You end up making some kind of magic, where somehow you end up doing something pretty good in a surprisingly narrow
amount of time.

Natalia 1:26:51 Let’s proceed to the last question. Do you think academia might change at some point into a more collaborative environment? I wish I knew that too. I’m curious about the answer.

Andre Marques 1:27:08 I think I’ll have to if it wants to survive. I think that a lot of areas, artificial neural networks, and artificial intelligence, in general, are good examples of a sector where state-of-the-art advances are increasingly coming from outside the ivory tower. People are still getting trains in there. And in neuroscience with the emergence of neurotechnology, engineering and programming are also going to start happening in other fields. I don’t know. But it’s gonna become a more attractive, less toxic, more merit-driven environment to survive, and it will survive because it’s been around for a long time.

Chances are that it will have to adapt. To end on a more positive note, one thing I would say is a lot of discussions that are happening a lot right now about open science, data sharing, open publication, and so on. These are things that were completed during my PhD. They weren’t nearly as discussed as they are now. I don’t remember having those types of discussions when I was working on my thesis. It wasn’t a topic that was thought of as important and discussed as much right now.

That shifted a lot in 10 years by luck. I didn’t exist back in the day. It’s a very slow and very conservative sector. Maybe, there’s hope.

Natalia 1:28:59 I couldn’t agree more. I think that if academia is interested in keeping the most people, then they have changed because they don’t have to keep them. Okay, I think that was the last question. So thank you. I would like to thank everyone. Andre, just say anything like your perspective, any piece of advice that you could give to young researchers like all these insightful things you gave us in this webinar?

Andre Marques 1:29:46 There’s one very specific piece of advice that I do want to share because someone gave it to me and I thought it was good. I would never have thought of going to recruitment agencies to get your profiles, your CVs. A lot of companies, especially big companies, their first screening that they do it through the CVs is exclusively through a big HR outsourcing group. And the way those companies work in HR is that they want to employ you if they think you’ve got an attractive profile. They get commission on placing you somewhere. If you build a good relationship with a particular network of recruiters, this can be a powerful way to branch out your profiles to many companies.

In general, I would say, we tend to think of our potential as humans, as a fixed thing. You know, it’s sort of, like a glass of water, which can be more or less full. We tend to think of it as a bit static. You reach your potential or you don’t, and this is not true. You can get to depend so much on the mentorship that you’re getting in the environment. You are on the amount of self-confidence that you build, because of those two things. For the football fans, or the sports fans in the audience, think about Moneyball, think about winning the Champions League with a bunch of people who could barely run in a football club. When you think about this, you start realizing, it’s not all up to you and that’s a bit frustrating.

It means you have to place yourself in the right environment with the right people and have a good mentor. You need to have a good network of people to learn and get inspiration from. The other thing it means is that no one can tell you what you can do or not. I genuinely believe that sky is the limit.

Natalia 1:32:23 Okay. Thank you, everyone. And thanks Andre your great insights. Thank you everyone for participating. If you still have some questions for Andre, he will be happy to take on these questions. Don’t hesitate to reach out also on LinkedIn and just poke us. Connect with us. That’s one and most networks already. Take the opportunity. And thank you so much for a very deep conversation today. I hope to see some of the locals will now be also traveling for careers to LinkedIn. I bet 100 euros even today that I will see you as co-founder of your company. Let’s see. I want to see if I was right. Thank you so much. Thank you everyone and have a nice evening.

Andre Marques 1:33:53 Thank you.

Leave a Comment