SDS 149: Data Science Tips for Startups to Large Companies

SDS 149: Data Science Tips for Startups to Large Companies

data science tips for startupsWelcome to episode #149 of the Super Data Science Podcast. Here we go!

Is artificial intelligence hype worth it? Is it valuable in the future? Listen to Jason Widjaja, the Associate Director of Data Science for Merck and Co., as he shares his experience in the field of data science and also discuss the future of artificial intelligence.

Subscribe on iTunesStitcher Radio or TuneIn

About Jason Widjaja

Jason Widjaja is the Associate Director of Data Science for one of the largest pharmaceutical company, Merck and Co. Jason is well-versed in the field of analytics, artificial intelligence, and technology. He advocates the correct and ethical usage of AI in different industries whenever needed, and also mentors and imparts goal-driven strategies to startups.

Overview

Listen to this Super Data Science Podcast as Jason Widjaja gives us a peak on his career journey that made him become a successful data science leader. He then shares his insights on how to progress as an effective data scientist lead in an industry with an entirely different setting.

For those who are considering career jump or career shift, learn how to meet the gaps of doing data science in a different industry as he familiarizes you with the intricacies of industry domain knowledge, functional domain knowledge, and data science domain knowledge. Jason attests that to excel in data science leadership, one must be a great team player by initiating collaboration and understanding every colleague’s domain.

So what really is and could be the value of Artificial Intelligence right now in businesses? Listen as Jason answers the question and enumerates progressive areas and industries that have proven that embarking on artificial intelligence has good payoff today and also in near future.

Jason also gives insightful tips and strategies for entrepreneurs on how to successfully kickstart startups in the field of Artificial Intelligence. He gave the following pieces of advice for startups:

  • unlearn the corporate methods
  • put/present your ‘self’ in front of customers fast
  • don’t be afraid to present unfinished products for feedbacks
  • find a co-founder that has complementing skills with yours, and lastly
  • look at conference papers to know the time gap and to not miss opportunities to jump in before corporate companies do.

Jason will also share his experiences when he built a startup– what to do and not to do especially if dealing with artificial intelligence.

There are a lot of opportunities for AI-inclined startups out there! Jason will guide you point-by-point as you decide whether to choose B2B or B2C and what type of industry to look into while taking in mind your vision and goals. At the end of this Super Data Science podcast, Jason will discuss the importance of having an ethics unit in an AI-centered industry. Ethical use can easily prevent systematic discrimination, bias data, and prevent heavy regulations to hit your company.

Tune in now for more!

In this episode you will learn:

  • Jason Widjaja shares his experience working as a data scientist lead for a big pharmaceutical company. (7:10)
  • How do you become a successful data scientist shifting to an industry you know nothing about? (12:25)
  • The distinction among Industry Domain Knowledge, Functional Domain Knowledge, and Data Science Knowledge. (14:38)
  • The different types of analytics – descriptive, predictive, and prescriptive. (23:30)
  • Is the world finally moving in the direction of Artificial Intelligence?(27:20)
  • How do you build successful startups venturing into the space of Artificial Intelligence and Automated Data Science? (29:32)
  • What type of startup do you choose for AI – B2B or B2C and what industries are considered lucrative in the space of AI? (39:35)
  • More progressive countries have considered the use of AI in businesses. (48:00)
  • How important is practicing ethical use of AI in businesses? (53:45)

Items mentioned in this podcast:

Follow Jason

Episode Transcript

0

Full Podcast Transcript

Expand to view full transcript

Kirill Eremenko: This is Episode #149, with Associate Director of Data Science at Merck and Co., Jason Widjaja.
Kirill Eremenko: Welcome to the Super Data Science Podcast, my name is Kirill Eremenko, data science coach and lifestyle entrepreneur. And each week we bring you inspiring people and ideas to help you build your successful career in data science. Thanks for being here today, and now let's make the complex simple.
Kirill Eremenko: Welcome back to the Super Data Science Podcast ladies and gentlemen. Today we've got a very exciting guest, Jason Widjaja, on the show. Jason is an Associate Director of Data Science at Merck and Co., which is one of the largest pharmaceutical companies in the world. In 2017, they had 40 billion dollars of revenue and they employ about 69,000 people.
Kirill Eremenko: So Jason is an Associate Director of Data Science and he's based in Singapore. He runs a team of 12 data scientists, and moreover, he's a very well versed person in the space of analytics, artificial intelligence, technology, and data science careers. He's a top contributor on Quora in the space of analytics and today he came on our show to spend some time and share his experience and knowledge.
Kirill Eremenko: So today we'll talk about his path into data science, what background he has, and how that led him into the space. Then we'll see what it is like to be a leader in the space of data science in a large company. And Jason will provide his own tips to those starting out into the space of data science looking to acquire certain knowledge and become proficient. And we'll also talk about interesting things such as industry domain knowledge, functional domain knowledge, data science domain knowledge, and so on.
Kirill Eremenko: But we don't stop there, Jason has worked quite a bit with startups in space of artificial intelligence advising people, helping people out, helping companies out, understand how to vary apply technology. And Jason will share a lot of his experience for those of you who are looking to get started in the space of artificial intelligence, set up a company, and change the world through AI. And even if you're not looking at that yet, I think it's a good overview of what's going on in the AI startup space, what's the different between hype in AI and actual value and how that industry can change the world.
Kirill Eremenko: And finally at the end of the podcast we'll talk about ethics in artificial intelligence. So, as you can see, this podcast is packed with value, lots of different topics are going to be covered, lots of different advice. I really enjoyed our conversation and can't wait for you to check it out as well. Let's dive straight into it. Without a further ado I bring to you Jason Widjaja, Associate Director of Data Science at Merck and Co.
Kirill Eremenko: Welcome ladies and gentlemen to the Super Data Science Podcast, and I've got an interesting and exciting guest with us, Jason Widjaja, who is an Associate Director at MSD, an artificial intelligence and data science products company.
Kirill Eremenko: Jason, welcome to the show, so excited to have you today. How are you going?
Jason Widjaja: Hi everyone, really good, thanks. And hi from Singapore.
Kirill Eremenko: That's awesome. So, Singapore, it's like- I'm in Brisbane right now, two hours difference as we discussed, and yeah. How's the weather again there? You said hot and humid, right?
Jason Widjaja: Yeah, hot and humid kind of describes probably about 366 days a year. The only variation is hot and humid or hot and humid and wet, and today it's just hot and humid.
Kirill Eremenko: Okay. I saw this one video where some marketing company was really smart, I think they did this in Singapore, what they did is they put invisible painted stickers on the pavement, on the sidewalks of the roads and then they're invisible, but when it rains, water kind of jumps off them, so they are completely dry or something, so you can see them in the rain. So you're walking and you see them. Is that in Singapore? Have you ever thought of that?
Jason Widjaja: I'm not sure if that's in Singapore, but I've not seen it. Maybe I just need to get out more.
Kirill Eremenko: Yeah, but I thought it was a cool idea. Yeah, alright, well. You are a good friend of Eugene duBossarsky, is that correct?
Jason Widjaja: Yeah, he's a great guy. I view him as a mentor of sorts and he certainly helped me out a lot as I was getting into the data science field.
Kirill Eremenko: Yeah. We had a chat with Eugene a couple months ago, he was on the show as well, on the podcast, and I totally agree. Very, very interesting person and I can totally see if you the opportunity to catch up with him, because you worked in Australia for seven years or something, something around there? I can totally see how he could become somebody's mentor.
Jason Widjaja: Yep, that's right. I was mostly in Melbourne, but whenever I get a chance up in Sydney I will try to catch up with Eugene. So, yeah. If anyone has not heard his podcast, I've heard it as well and it's pretty god, so go check it out.
Kirill Eremenko: Oh nice, that's nice, for sure. How do you guys know each other, with Eugene, just out of curiosity.
Jason Widjaja: It's really interesting, because I think I reached a stage that I was really looking for advice and mentorship, and this was probably about five years ago. And it was really interesting because I remember searching again and again in the data science communities of both Sydney and Melbourne and this guy's name just kept coming up again and again. So, I didn't know at the time, but he found data science Sydney and data science Melbourne, and our user group, i was just- my kind of tool of choice, and so I kept running into this name and one day I just decided to contact him out of the blue and said, "Hi, I'm in Melbourne, you're in Sidney, but I would love to catch up with you if you ever come to Melbourne." And it just so happened that he was in Melbourne that very afternoon and we met within hours that note and we've been meeting up ever since.
Kirill Eremenko: Wow, that's so cool. It's like fate, right?
Jason Widjaja: Yeah, yeah.
Kirill Eremenko: Had you messaged him a week later, it probably wouldn't have happened.
Jason Widjaja: Yep, suddenly you're one of the most interesting people I know.
Kirill Eremenko: That's fantastic. And just for listeners, if you want to check out the podcast with Eugene, it's Episode #131, we talked about the purpose to data science and the truth about analytics.
Kirill Eremenko: Okay, but coming back to our podcast today, I'm very excited about the things we will talk about. Just for the bit of our listeners, guys, Jason has some very impressive qualifications and a very deep and diverse at the same time background, both in business, and AI, and technology. There's lots of things I'd love to talk about, I don't know how much we will fit into this podcast. But tell us a little bit about yourself. How did you get into the space of data science in the first place, Jason?
Jason Widjaja: That's a quite interesting one as well, because I kind of stumbled into it, rather than explicitly going out looking a role in data science. Just to share, my background is kind of undirected. So I eventually went back to university three times. My first degree is in IT. Then that doesn't quite hit the spot, so I did my honors in human resource. Then a few years later I did MBA, and I still was not getting it, so I eventually went back to do my second master's in analytics. And that was mostly to pursue my interest in machine learning and LP. So I kind of did these qualifications with five year gaps in between.
Jason Widjaja: And it was only towards the end of my final qualification and after all this time the recruiters were looking at me like, "Jason, you are undirected. You're middle aged, make up your mind and do something."
Kirill Eremenko: If you're an eternal student you just keep studying all the time.
Jason Widjaja: Yeah, you just keep studying. You know, are you going to get off your- get out of school and do something worthwhile? For society and that kind of stuff.
Jason Widjaja: Then around that time it was in the maybe 2013, 14, people would start saying, "Oh, you have two [inaudible 00:08:49] Venn diagram, and you have business stats, and computing?" And people were saying, "You need to find unicorn, but it's too hard because you need to do it as a team because no one will have all these three qualifications." And I'm like, "Well, wait a minute. Actually, my three degrees kind of cover all three of those." And I'm just like, "Yay, pick me, pick me."
Kirill Eremenko: "I'm the unicorn you're looking for," right?
Jason Widjaja: Yeah, yeah. Slightly overweight Asian unicorn. But yeah, suddenly I went from, "Make up your mind" to, "You're really important, you're really wanted."
Jason Widjaja: I think just to say one thing about the Venn diagram, I think it is some, I will call it not complete because people come into data science for many, many skills, and if you were only to look at stats degrees and programming degrees, you're missing out lots of good talent. For instance, in the company I work for, MSD is actually a pharmaceutical company, and so you have this entire large group of scientists. So it's not data science starting from data, it's data science coming from the scientific method. And those people have neither stat degrees, nor computing degrees, but many of them are really good in data science. Just something about that Venn diagram before we get to an [inaudible 00:10:11] of it.
Kirill Eremenko: Yeah, no, totally agree. I think it's more of like a general type of filter I guess for recruiters or a general sense of direction, but indeed they are unique cases or industry specific cases when you have people who might not have all those three components, but might be still good data scientists. Yeah, totally agree with that. It's just a good way for somebody to understand the complexity of data science, but not necessarily an exhaustive [crosstalk 00:10:46], or the only criteria.
Kirill Eremenko: Alright, cool. So right after uni, after your third or fourth degree, you got into MSD, into the pharmaceuticals, is that right?
Jason Widjaja: Yep. So I've only been with MSD just for over two years and I joined the data science team here in the Singapore hub. What I see when I get to work is we have probably approximately 300 technology professionals in one place, and it's a really interesting environment because there is a very big gap between doing data science on your machine and talking to a client and doing data science in production and for the later case, we can add all other parts of IT as well. I work in I guess an IT hub kind of environment.
Kirill Eremenko: Gotcha, gotcha, okay. And so what would you say you spend more time doing, data science or tech?
Jason Widjaja: I think you spend your time doing data science, but coordinating with tech because I think we don't really want data scientists to spend all that- after getting, spending X years to pick up all these skills and stats and machine learning, you probably don't want them doing just software engineering or use [inaudible 00:12:08] or testing because there are already massive IT teams who are specialists in these areas and they can do it better than us. My simple response is generally data science or data scientist.
Kirill Eremenko: Yeah, okay. Very interesting career paths that- as I understood you never studied pharmaceuticals before getting into that space? Is that correct?
Jason Widjaja: That's correct. That's something that's stares me in the face every day. Yeah, yeah.
Kirill Eremenko: Was it hard to pick up in a company- because I think that's a great example of somebody studying degrees related to data science or [inaudible 00:12:52] in the space of data science and then bam, you jump into pharmaceuticals, which is a completely foreign type of industry, and yet you are able to be successful. So can you tell us a bit about that journey? What did it take, and moreover, you started as a senior data scientist and then six months later you are now an associate director in global data science, that's a big jump in terms of career. What's the secret to your success, what's the secret source to be successful in an industry that you know nothing about?
Jason Widjaja: I think there are many lessons at play here, because we do have the industry and pharmaceutical is on lens. And when someone mentions pharmaceuticals I'm not sure what people may think about, but they may think about proteins, and drug discovery, and genomics, and so on. But there's also a functional lens, where every company, whether you are pharmaceutical company, or mining company, or ecommerce company, you have sales, you have marketing, you have IT, you have HR, you might have a plan, you might have finance. And so, it was quite jump, but also at the same time, not a big jump because although I didn't have the industry domain knowledge, I do have quite a lot of functional domain knowledge. And that is a good way for people to switch industries.
Jason Widjaja: For instance, if you do sales and marketing in one company, you can transfer lots of those skills to a different industry to do sales and marketing in a different industry. And the theme for any of the [inaudible 00:14:28]. So it was not that bad a jump. But having said that, I am still spending my time learning the nuances of the pharmaceutical industry.
Kirill Eremenko: Gotcha. Okay, that's a very interesting comment. I haven't heard that term before, functional domain knowledge, because you're right, when you think pharmaceuticals, you start thinking about drugs, you start thinking about genome, and stuff like that. It's very interesting to hear this perspective that in reality, functional domain knowledge is very transferable, right? It's a transferable skill. If you know how to do marketing in one company, yes, there are going to be specifics in different industry, but at the same time, you're got 80 or 90 percent of the skill set that you have, running a logistic regression, or using R and Python in a certain way, you're going to be able to apply them in a different industry. That's very exciting to hear.
Kirill Eremenko: You mentioned that you're still learning the pharmaceutical specifics, why is that? Is that to augment your role even further or are there inevitable requirements that you can't get around with just the functional domain knowledge?
Jason Widjaja: I think a bit of both. In terms of learning the other functions that I'm not familiar with, I think that it's more to not waste the time of my colleagues when we need to work together across teams. So something I see at work is that we often go up in silos, so when you take a major, in a degree, you're basically going up in a silo. When you work in department you're going up in the silo. But I think business happens kind of all at once. I kind of advocate as much cross functional collaboration as possible. And to do that we need to get out of our bubbles and shells and understand other people's domains. And for me, that means starting to understand a little bit about what we call [inaudible 00:16:32] evidence. Or things like, bioinformatics just had a- at least a cursory level so I understand what my colleagues are doing.
Kirill Eremenko: Mm-hmm (affirmative)- gotcha. No, totally agree. I like the way you put it that you don't want to waste your colleagues time. I think everybody appreciates that when their time is valued and not wasted. They will see that you're making that effort. Definitely some great advice here, especially for those data scientists who are moving to industries that they haven't worked in before. Some interesting tips here.
Kirill Eremenko: Yeah, sure.
Jason Widjaja: Just to quickly add something on domain specific knowledge. So you have the industry, you have the function, but if your hiring manager is not a data scientist, you actually have one more layer as well, which is your data science domain. For instance, if you could be a specialist in natural image processing. And NLP lends itself to use cases in- across functions, so it be NLP person for HR, it can be NLP for IT, NLP for sales and marketing. It can computer vision for manufacturing or computer vision for facilities. I think there are many ways you can shape yourself to fit in different niches depending on what you want to do and who you're speaking to.
Kirill Eremenko: Okay, thank you for the tip. To kind of outline that, we've got the industry domain, the functional domain, and the data science domain, they overlap. It's very interesting because when I say "domain knowledge," I usually imply just the industry domain knowledge, but I guess that's a good clarification that it's not the only domain. It's better probably to say industry specific domain knowledge, or business knowledge. And then the whole of the domain knowledge can be broken down into these three components.
Kirill Eremenko: Alright, well thank you for that. Can you tell us a bit about the move? That move from senior data scientist to associate director, what is your role currently entail and how did that happen. As I see from LinkedIn, it's happened pretty quickly, just in six months.
Jason Widjaja: Alright, so I think the six months was more of an extended probation period for me because I had to cross over that industry hurdle as well as managing a team that is larger than the teams I'm used to managing. So currently I lead a team of data scientists, currently about, I think, 12. By the middle of the year it will probably be around 15. Something we try to do is, like I mentioned earlier, data science for data scientists. So when we bring in data scientists they concentrate on prototyping, modeling, and we try to stay away a little bit more from things like data management, reporting, and business intelligence.
Jason Widjaja: My current role is, there's a large people management portion of it where we have this 12 professionals who really takes up most my life actually. The other bit is just leading the team and when I say leading the team, I think it's trying to shape the environment in a large corporate to make it conducive for data science. Because in many ways I think many large corporates, they are used to up front planning, long term business processes, and financial processes, and business cases, and so on, which is sometimes at odds with the nature of data science. We are trying to navigate answers and test use cases out, see if data quality stands up to a certain kind of application, and so on. So I try very much to I guess create an environment where data science can strive. And I think that's probably the other big thing that I do.
Kirill Eremenko: Yeah, no, that's a very interesting way of describing data science leadership. I think I encountered a similar thing when I just left Deloittes and I went into the finance industry. Of course not at the same scale, I was only building a data science team of one, which was me, and then we got another person on board. I feel I can relate to that, that data science leadership, or being in charge of a department is more kind of- a big component of that is talking to other parts of the company so that they're aware of what work you do and how you can help. And also that they're more open to helping you actually do your job and not blocking access to data and they're just treating you as an internal consulting component of the business, but at the same time, also actively helping out you in your role. Would you say that's kind of the main parts? The main path that you lay for the team?
Jason Widjaja: Yeah, I think you describe it very well. I think that the model of an internet consultant is basically what we do. So I'll contrast that with a bit of service provider, where we invite ecommerce, we get requirements, and we build solutions. Often I think we are not at the stage where the broader business community has a clear idea of what data science can and cannot do and how they should interact with data scientists. So, often when it comes to shaping the projects and shaping what we call products as well, we do need to work to get out front with the business and kind of follow them along the way as well. Because I guess what we encounter is the business clients are often very comfortable with asking for say, a mobile application, but they are less comfortable with what it may, what the data science deliver might look like. And we do get requirements that are quite vague at times and we just need to not take these wholesale, but say what would the business, can we boil it down into something that is possible, feasible, and valuable.
Kirill Eremenko: Gotcha. No, totally agree. Okay, and so that's your current role. When we're talking or actually when you wrote about what you'd like to talk about on the podcast you mentioned that you have a focus on artificial intelligence and automation side of data science. So is that in your current role, or is that more of hobby?
Jason Widjaja: So I think that to clarify, I lead one of the data science teams, I offer two teams in the region. To take a step back we basically have one of these slash IT hubs of a few hundred people across each of the, I guess I'll say time zones, one in the US, one in Europe, and one Asia/Pacific and basically I'm in the one in the Asia/Pacific. We have two teams, one team we, I guess the way it's evolving, we would kind of call it decision science. So, data science in support of business decisions.
Jason Widjaja: The other team, which is my team, we are evolving it into something that- we're still deciding on names, but basically it's a focus on AI and data products. And the tension between the two is quite interesting one. It is who makes the decision. So generally when we encounter like a white papers talking about analytics we have this majority model of a descriptive, and predictive, and prescriptive. I think I will add that the entire model kind of presumes that it is analytics for a human that's making the decision.
Jason Widjaja: But I think the way we would challenge it is, well, if they're prescriptive analytics, like prescriptive implies the software actually knows what is the best decision, and if the cost of a bad decision is negligible and pretty good is good enough, then we can actually have autonomous execution authorization as well. And that's where it kind of flips over to AI.
Jason Widjaja: So the reason why we kind of have automation as a bit of a subheading is because the skills to deploy AI products is a super set of the skills that it takes to deploy automation solutions as well. So, that's something that we kind of do not with a focus on, but as part of the different things we do, we might do automation as well. Our focus is really on AI and data products.
Kirill Eremenko: Interesting, okay. So just for the benefit of our listeners, if you're not familiar with the model, is it correct that it's the Gardner Model? That descriptive, predictive, prescriptive?
Jason Widjaja: I am not so sure, I can look it up.
Kirill Eremenko: No, that's okay, I just know that it came up originally, where I originally encountered it was in Gardner and we'll include a chart in the show notes and basically descriptive is what happens, predictive is what will happen, and prescriptive analytics is how can we influence the outcomes, or prevent it from happening, or actually make it happen in future more.
Kirill Eremenko: And so the decision support side of data science is different stages in that process, right? It can either help with just the descriptive, or the predictive, or the prescriptive parts of analytics. Whereas AI, which you're actually saying is you take of that and you take it the next level and you get the artificial intelligence to makes those decisions on behalf of humans. That's very interesting. I don't want to go into too much detail because I'm sure there's a lot of trade secrets and stuff that definitely you cannot disclose, but that's an interesting move.
Kirill Eremenko: Do you think that overall that the whole world is slowly moving in that direction or is it still in the phase of companies are still testing the waters, seeing if this is possible?
Jason Widjaja: I think there's a place to make to a distinction between the AI in terms of the marketing and AI in terms of what's actually being implemented. So, I think something I do struggle with at the moment is that AI is kind of at the- if you know the hype cycle, AI is pretty much near the top of the hype cycle at the moment. So, understand if you are a startup and you have AI in it, your valuation goes up a notch.
Kirill Eremenko: So sure, yeah. No, it's more like 10 times.
Jason Widjaja: Yeah.
Kirill Eremenko: You put in the words AI and blockchain. [crosstalk 00:28:11].
Jason Widjaja: Kind of use them both together, yeah.
Kirill Eremenko: Yeah. I feel like sometimes people, with these ICOs that are happening, people just do ICOs for the sake of doing an ICO and having that word blockchain in the name of their style because they they did a nice [crosstalk 00:28:26]. But sometimes there's not even a point to having that cryptic currency tokens supporting your idea because you're not ever going to use it. It's just like, "We did an ICO and raised some money and that's it."
Kirill Eremenko: Yeah, but sidetracking on that. You're right, it's still in a hype kind of phase. Do you think there is value to having AI in our organizations?
Jason Widjaja: Yeah, I think [inaudible 00:28:55] and I guess I have an obviously bias here because that's like the work that occupies most of the time of our team, but I think [inaudible 00:29:02] case for AI is very good. But it's also something that must be done with the other business community in mind. Because it's quite easy for AI to cause harm, just in the journey of being very efficient at what it does. So we can talk a bit about that a bit later. I think certainly the business case for AI is very good.
Kirill Eremenko: Yeah, no, I totally agree. Okay. This actually brings us to an interesting point, that you actually, apart from your main role at MSD, at the pharmaceutical's company, you also are passionate about working with startups which are in the space of artificial intelligence and automated data science. Can you tell us a bit more about that? I don't think we've had the guest on the podcast who helped out with some tips and ideas for people in the space artificial intelligence who are looking to get started, who are creating a startup, who might have an idea. Maybe you can share some of your experience with working with startups and what kind of mistakes they normally make, what is the journey like, what are the requirements. What could you share that could benefit somebody who's at the start of this journey, or even just considering creating an artificial intelligence startup?
Jason Widjaja: I think I'll start with something that will apply to most startups and I'll tell you where I'm coming from as well. So, when I was in Australia, one of the things that I did was I was a judge for business plan competition and over the years as I saw more and more young companies I jumped into a couple of them myself I tried to do one, two startups and I failed miserably. Basically why I share is kind of lessons learned from the game, my own ones, and also from looking at probably a few hundred companies, mostly in the tech space. And I think it depends on where you're coming from when you do your startup. But I think one really common mistake is to take the professional large corporate mindset and put that over into startup world. So, that world looks like having value prepositions, and documentation, and plans, and wire frames, and a whole lot of up from work before a product actually reaches it's customer.
Jason Widjaja: That's probably one of the biggest reflects I encounter. And I see this mostly when we have more experienced professionals crossing over into startup world, perhaps with their savings from the first few years of work, or so on. I think that doesn't work at all. And basically what you want to do is, if you want to pick up one book, I'll probably suggest The Lean Startup. I think that mindset is very useful, it's almost a ready, fire, aim, ready, fire, aim kind of mindset.
Jason Widjaja: And I think having the thick skin to put very raw things in front of customers as early as possible is very, very valuable and something that is not done very much in larger companies. So if you're coming from some professional department in a large company- well I think one insight is [inaudible 00:32:43] structure of large company and [inaudible 00:32:45] structure of a small company, of small startups is exactly the same except that your name's against every single role instead of the large company.
Jason Widjaja: So, suddenly there'll be whole world of things that you don't know how to do. And you don't need to know how to do to a large extent, you just need to know enough to survive. So you've gone from being an expert in very narrow domain, to finding your feet in a dozen different unfamiliar domains at once. And in the midst of that you still need to get your product in front of customers as soon as possible. So it's very easy to go under. But yeah, I think I'm learning most of the big corporate mindsets is definitely a good lesson. Having the lead startup, put yourself in front of the customer as soon as possible, is also a very useful lesson.
Jason Widjaja: And I think the last one is probably the most difficult thing, I find personally, is to work with someone who is very different from myself. And yet I think that is probably crucial to success because when you have a team of like, two, between two of you you need to cover the entire world of your business. So you better make sure that this person has a complimentary skillset to you. And often that is not easy because we tend to naturally gravitate towards people who are more similar to us than different.
Kirill Eremenko: Mm-hmm (affirmative). Okay, that's very cool, and so let me just sum those up. I'm learning the corporate methods where you wait a long time before you present the product or you feel it's ready and you have a big company working and everything. And then pull on from that, putting yourself in front of the customer fast, and not being afraid to iterate and show unfinished products and get feedback. And finally, you need complimenting skills with your co-founder or co-founders so that you guys double up on different strengths rather than having the same strengths.
Jason Widjaja: Yeah, yeah. That's a great way to summarize it. And I'll just add one AI specific one. AI is really interesting because unlink many other types of startups, many other types of startups are not working from the HR for research. So in AI, the wonderful thing is we have this ritual of published papers often are implemented court. Usually it takes a number of years before the cutting edge papers maybe you see at a NIPS conference and IPS before it makes it into a large enterprise tool. So that space is a space that you have opportunity and I do see quite a few AI startups simply implementing the right papers from conferences before a large corporate is able to. I think the open source will, and the world of implemented court from conferences is such a rich space to mind for cutting edge technology.
Jason Widjaja: We used quite a lot of data ourself at work and so we sometimes look at the repos of Facebook and Google and we actually fox some of those for our own projects as well. That is, I guess the joy of being part of the open source community and it's really a rich source of AI technology. Spend lots of time looking around conference papers for the tech you need.
Kirill Eremenko: That's a great tip. So, very AI specific to look at the papers that come out in conferences and you have a time gap before a large corporate will implement it, so in that time gap you can quickly create a startup around [crosstalk 00:36:48].
Jason Widjaja: Yep, true story.
Kirill Eremenko: Yeah, awesome. I just wanted to see, are you comfortable to talk a little bit about your startup and why it failed? I'd love to know. But if that's okay with you, if not we can move on to something else.
Jason Widjaja: Yeah, sure, no problem.
Kirill Eremenko: Just like in short. I think it might be useful to listeners to learn from your experience in that way.
Jason Widjaja: Yeah, I think something I did wrong was to firstly not do development in house. At that time I was somewhat technical, but not good enough to create a whole software program myself. In the interest of saving a few bucks up front, and it was my first time going out, I basically outsourced to an overseas provider. Since then, I've always encouraged you to have developers in house because, I think a lesson I took from then is that words are a terrible medium to convey technology requirements. I used to write these long emails and long multiple page documents, five to 10 pages of different features and different requirements and so on. I think I wasted a lot of time doing that when I could have just knocked up a very simple, basic wire frames and prototypes as opposed to writing long documents. So yeah, documents for requirements didn't work for me.
Kirill Eremenko: Mm-hmm (affirmative). Yeah, I bet. So it kind of slowed down the development process I guess, or made it harder to get exactly you want from the developers.
Jason Widjaja: Yeah, yeah
Kirill Eremenko: Okay gotcha. That's some great advice. Alright. The space of AI is huge, right? I like Andrew Ng's quote that, "AI is the new electricity" and basically if you look at a hundred years ago back at the start of the 20th century, electricity wasn't common in businesses. If you look now, you cannot even name a single business that doesn't use electricity. Like, any business, however remote it is and anything, everybody uses electricity. Same thing goes for AI. Like now, it's not ubiquitous, it's like a couple of companies here and there, leaders and pioneers are using artificial intelligence already. But I think everybody kind of expects is going to happen much faster, it's not going a to take a hundred years, it's going to take 10, 20, 30 before AI is everywhere.
Kirill Eremenko: So, while that's the case, while we're still in this developmental phase and companies are jumping onboard, I wanted to get your views, or from your experience, what would you say is the easiest place, or the most common type of startup that's working in the space of AI? Let's start with, is it B to B? Business to business type of startup? Or is it B to C, business to customer type of startup? And then if you can expand a little bit, what kind of industries do you foresee the most lucrative for AI startups in the next two or three years.
Jason Widjaja: Right, so I think there's a few things there, so let's try to unpack it. I think there's one component of AI being everywhere and that's the AI that will not surprise us because it's early working behind the scenes to power so many things that we take for granted. For instance, when we use Google Translate, there is AI technology underpinning that. When we go to any social media feed we have new feed recommendations and prioritizations, which kind of happen seamlessly, although it would be some similar technologies powering that as well.
Jason Widjaja: And so from a startup perspective, almost every startup I see nowadays has some mention of machine learning or AI in it. Although if I were to be pretty strick in academic way, I think lots of it is not the case. And just seeing it for the sake of having a me too and not to lose out from a marketing sense.
Jason Widjaja: As for the dedicated AI startups, they also come in different flavors. And it really depends on who your customer is. Usually we see a lot more B to C startups as opposed to B to B or B to E. Just as a general truism of any startup ecosystem. So, it's the same here in Singapore and it was the same when I was looking around in Australia that B to C startups outnumbered B to B startups by at least maybe four or five to one.
Jason Widjaja: I not sure why that's the case, but I think it's just that many startup founders are people who are a bit younger, and so they may not have had many years of professional experience to get their minds in to the B to B role. Most startups tend to be retail startups, consumer startups, social startups, platform startups, two sided market startups, and so on.
Jason Widjaja: Whereas from an opportunity perspective I would generally lean towards the B to B or B to E startups simply because number one, there is less competition. Number two, there is a ready stream of tech that is available for you to start to draw on. Number three, I think corporates, at least from my limited symbol set, are starting to take the whole world off. They've [inaudible 00:42:49] AI machine learning seriously enough that they are starting to have enough budget to try new things with startups. So I think startups are enjoying a more wonderful opportunity over these few years to lend some big corporate clients to starting from pretty simple [inaudible 00:43:09] that are within their reach.
Jason Widjaja: So in general, those would be the directions I would aim for at the start.
Kirill Eremenko: Mm-hmm (affirmative). Totally, I see that works. I'd like to add to that that I've already mentioned before on the podcast that my business partner, Hadelin, we create courses together, and we actually started our own startup just a few weeks ago, maybe a few months now in artificial intelligence. But, we did that in the space of B to B. It took us a lot of brainstorming last year, in the summer last year, we literally spent a whole month brainstorming what to you. We see the value of artificial intelligence, we see the way it's going and we want to contribute to the world, help however we can.
Kirill Eremenko: And it's interesting, it's a fun space to be in. And at the start we were thinking a lot of B to C, B to C, impact a lot of people, help millions of people through artificial intelligence. It sounds very noble and sounds like a great way to go about it, but then we, at one point I remember we were having lunch in Paris, and we completely shifted our view because as you say, the competition in the B to C space is massive. I'm interested to hear the statistics four to one, B to C, versus B to B.
Kirill Eremenko: So competition is massive in the space of B to C, everybody's trying it out. It's much scarier or it's less easy to understand how to get into the B to B space. So I agree, that's another reason why people would avoid it and people are starting companies.
Kirill Eremenko: But also one other thing you mentioned really resonated with me, and that's the budgets. The corporates and the businesses and enterprises are at that stage now that they see the value and they see their own competitors in their industries taking on artificial intelligence and putting that competitive pressure on them. And so the ones that are smart, the companies that are smart are going to be considering artificial intelligence and allocating budgets because they know that it's an investment to their future and to their survival.
Kirill Eremenko: That's what we're seeing, that for companies it is- for an individual to see value from artificial intelligence and then reward the company that's actually providing that artificial intelligence with investment is quite difficult to demonstrate that because people, Facebook already has artificial intelligence. Most of the things that we use, like Amazon and so on, those companies are already powering it with artificial intelligence.
Kirill Eremenko: Whereas in the business world, there's so many business out there that are hungry for AI they don't have companies that can help then with AI and yet they are willing to provide substantial investments into that space because in the long run it will pay off tenfold, hundredfold, thousandfold. Or it will just basically allow them to survive.
Kirill Eremenko: I think for those listening it's a very interesting thing to consider that even though the B to B space might sound scarier and might sound bit harder to grasp how to get into it, at the same time the rewards there are much more substantial. In my perspective, and in our assessment, rewards are much more substantial. And the impact you can make is also much more substantial because even though you can probably help a million or a billion people, because your B to B, not B to C, but at the same time you can help. Even if you help like, five businesses, that can be a massive, massive impact.
Jason Widjaja: Totally agree with everything you've said. The benefit of going down the B to B route is that there being only a few businesses, a few [inaudible 00:47:14] industry actually works in your favor very, very well because use cases, they are quite specific to industry. If one competitor sees another competitor doing it, the next person would almost automatically be like, "Who did that for you?"
Kirill Eremenko: Yeah.
Jason Widjaja: And that's a really valid question because I think for big companies who adopt data science or AI, it's a multi-year journey. And if you start when you already see other companies reaping the results it might be too late. So I think companies who are a bit more forward thinking are already in that frame of starting to experiment and that is great if you are an AI startup.
Kirill Eremenko: Yep, yep. Totally agree. Form your experience working in different regions, like you've worked in Australia, you worked in Singapore, and I'm assuming you have some connections or experience in other areas as well. Where would you say is the- what are the most progressive areas of the world where artificial intelligence is really starting to pick up these days?
Jason Widjaja: Oh yeah, that's really interesting because I think the answer to that question evolved a bit over the past few years. I remember having a discussion back when I was doing my second masters and we were like, alright, so we have the US and we have DeepMind in the UK and we have basically the big Chinese players. And the year after I think I [inaudible 00:48:42] quiet DeepMind, so it became from three players down to two.
Jason Widjaja: And so I've just back from Beijing, literally a few weeks ago. It was my first time to China, and I was really blown away by, at least in the major city of Beijing, how advanced they were and the mechanisms they had to scale technology up. So I think we're seeing an arm race between the US and China in terms of AI. But I think the mechanisms they use to go out doing is very different. Because China has a few large bit companies with central control and so on. Whereas the US has a- the country itself is set up so differently and so the landscape is quite different as well.
Jason Widjaja: One other thing I would add is, maybe two other things, is with the advent of Deep Learning we have a big pocket in Canada as well where you have Jeff Hinton, and [inaudible 00:49:45], and the [inaudible 00:49:47] of Toronto.
Kirill Eremenko: Montreal as well.
Jason Widjaja: Montreal, yeah. And so I think it's probably an investment offer some couple of hundred of million in that space. On the back off Jeff Hinton and his community over there. And so that's a very interesting spot to watch as well.
Jason Widjaja: In a totally biased way I came back in Singapore also partially because Singapore is trying to position itself as AI hot- in Southeast Asia. And so there was quite a lot of central government support for AI across industries from big corporates down to SMEs as well, and startups as well.
Jason Widjaja: So as a bit of disclosure, my younger brother runs an AI startup and he kind of chose to do it in Singapore because he saw how the government was helping us and the startups. So it's too early to tell how effective all that will be, but I'm still staying here for a couple of years to give that a try.
Kirill Eremenko: Mm-hmm (affirmative). Well congratulations to your brother, hope it goes well. It sounds like an exciting time for him. And another space, actually interesting enough, out of all the places in the world, Switzerland. Switzerland also has some AI startups popping up. I'm looking at a website of one of them, it's called NNAISENSE, like in the birth of AI or something like that. Led by people who are behind the LSDMs. LSDMs originally came from Germany/Switzerland, so Juergen Schmidhuber and others are on that project, and Sepp Hochreiter. And so yeah, I think it's because they have the university, which is a famous university in Zurich, that, I believe it's in Zurich, that Einstein went to. And they have some very strong capabilities there as well. Like the University of Montreal as well.
Jason Widjaja: Right, right.
Kirill Eremenko: It's around the university.
Kirill Eremenko: Okay, yeah, that's interesting. So, interesting spaces for artificial intelligence these days. I agree, it's really cool that a lot of the research is actually done through open research papers, isn't that right?
Jason Widjaja: Yes.
Kirill Eremenko: Like everybody can work- I don't know if the world's seen this type of advancement before, because usually if an industry like that pops up, you want to keep your things private and secret, and just leverage them yourself. But here we're moving into a world where companies like Tesla are just sharing their papers for everybody to use. And then you got the Google DeepMind doing crazy research, like cray great research, in the space of artificial intelligence, and all that is public. I mean, it's a world of opportunity for anybody who wants to get into that.
Jason Widjaja: Yes, yes. Maybe I've not been around long enough, but I've certainly not seen anything similar to this before. I'm not sure whether it's a case of we have the academic community just starting to overlap in terms of awareness with the broader commercial communities. Maybe the academic community has been doing this across many disciplines, but it's just that for some reason because of the size of the prize of AI people kind of wait for the commercialization perspective process to bring the tech to market. And people are just waiting at the door, or to university, so that, you know. Kind of getting it straight from the campuses.
Kirill Eremenko: Yeah, yeah, exactly. It almost sounds like we can create an A startup around that space, how to get the best talent from AI into your company.
Jason Widjaja: Yeah.
Kirill Eremenko: Versus from the university.
Jason Widjaja: Yeah, that's a B to B idea right there.
Kirill Eremenko: There you go, guys. There's an idea right there.
Kirill Eremenko: Alright, and so that's a very good discussion and I'm glad we had that. This could go on for a long time, but we're running out of time and I really wanted to touch on another topic that you are passionate about, ethics around the space of artificial intelligence. On this point, because you've already mentioned that AI can cause harm to the broader business community, let's talk about that for a little bit. Why are you passionate about ethics in the space of artificial intelligence and automation?
Jason Widjaja: Right. I think there are many different issues of ethics in AI. But I think once way to answer is just from a [inaudible 00:54:32] perspective when you're actually modeling your data and you're using machine learning on different data sets. You do see things that jump out at you and reflects [inaudible 00:54:43]. One stream of this is around systematic discrimination. And that's really around models that are trained on basically bias data. And in doing so the models are, they are simply the optimization machines and they systematize whatever exists in the current state.
Jason Widjaja: So as a couple of concrete examples, I think there was a research company off of Carnegie Mellon that said that it you are a male you see at ad for a high paying job more than if you are female. So somewhere down the line you have some sort of feature importance that's giving people points just by virtue of their gender.
Jason Widjaja: And the thing is, as more and more companies adopt these solutions without much thought of forensics or diagnostics, you just have a systematization of bias ad scale. So that's angle.
Jason Widjaja: But I think beyond that as well, there is also a dumbness in machinery models where there are basically, the models are what you feed them. And if you feed them based on current state, it almost forces you to come face to face with the state of society and ethics as it is currently. And if you don't step in to intervene to fix it at that point, you're basically pressing a button to say, "Alright, let's systematize whatever exists today and let's just accelerate it through artificial intelligence."
Kirill Eremenko: Mm-hmm (affirmative). Yeah, you're kind of taking it at unprecedented rates to an unprecedented scale never seen before and so you've got to be really careful about that.
Jason Widjaja: Yeah.
Kirill Eremenko: How are you going to do that?
Jason Widjaja: Yeah, and I guess it's a bit more of the data science perspective, but it's also the economic perspective as well. In the overlap after automations/augmentation space we do see the boundaries of what AI can do, how it's picking up different senses, how it's picking up different games, and so on. That boundary's being pushed pretty quickly. I think many, many, white collar process oriented tasks already fair game for AI and many more will be in the next- day by day. So this is other economic layer on top of this as well where you have what we call bounty [inaudible 00:57:27]. Where the majority of the benefits of AI would probably go to people who are already among the richest, small percentage of the population, and they'll accelerate as well.
Jason Widjaja: There is also the drop displacement angle where yes, AI creates jobs and yes, automation reduces jobs, but we have the problem where people with jobs are automated away can move into the the new jobs that are being created. And so we got a potential problem there as well.
Jason Widjaja: And then depending on where you are in the world, again, coming back from China recently I see the contract very starkly we have this whole world off data and privacy issues as well. This is like, massive tension between use old data in the name of security and hide all the data in the name of privacy.
Kirill Eremenko: Yeah.
Jason Widjaja: Where are on the spectrum each use case falls and if one person doesn't play nice, does it matter if everyone else plays nice. This is all data privacy issue on the issues of all this as well. So there are certainly many issues around just deploying AI models at scale and I think just to be a responsible corporate citizen and also to just build towards a world that you want to live in. It's important to have both the tech and business angle. Also the ethics angle, like moving along in step. Because I think you have BAs, you have business side, you have tech side, I don't really see a much ethics side of people in projects. And I think that could potentially be a problem if you don't address it at some point.
Kirill Eremenko: Interesting, okay. And so how do you think businesses can do that? Because the problem I see with that is that a lot of the things, anything a business does is either driven by necessity through regulation, or is driven by profit, by the bottom line. And that's just the nature of competitive markets, it's the nature of capitalism. And the question is, ethics, right? Having an ethic department or an AI ethics function in a business, does it really bring profit? No, probably no. And is it it regulated? Probably not as well. So, what is going to drive businesses to do that, because ultimately if you have five identical companies earning the same profit and then one of them decides to introduce ethics in artificial intelligence as a function in it's business, it's profit is going to drop because the costs are going to go up. It's costly to maintain that department, or division, or function. And so, what is the incentive for them to do that? How do you think the world is going to develop in order to accommodate this?
Jason Widjaja: Yeah, that's a really interesting one because I'm not sure whether companies are the best people to do it. It could be a case of what's it called? Tragedy of the commons where someone should do it, but if one party doesn't do it that party itself doesn't get penalized for not doing it. So I guess in one sense maybe the natural person to do these things would be not for profit organizations, could be government, could be think tanks, could be bodies outside of the commercial wall. But at the same time, I'm not sure whether being in commercial whether we should sit down and wait for a heavy hammer of regulations to come and hit us on the head.
Jason Widjaja: And I think it will do us good if we have as much self governess as possible. I guess I would prefer to have commercials take a light touch self governess as opposed to having a dozen new audit requirements, like hit on the head because you're just deploying a single AI project. So, yeah.
Kirill Eremenko: Okay, yeah. Gotcha. So it is an interesting approach to that as well. Kind of like, be prepared and when it does come you're ready for it rather than it's completely new that you get, as you say, hit on the head.
Jason Widjaja: I think our legislation and regulation will always lag practice by a number of years and what you do in a number of years depends on how you frame that situation.
Kirill Eremenko: Yeah, it's crazy with this couple of years whole situation. Because when you think about it, yes, indeed legislation lags by I don't know, two, three, five years, but that's if you think of it in linear terms. But if you think of it in the terms of the world we live in, which is exponential, five years is actually 32 years, yeah? Two to the power of five is 32. So, legislation lagging by five years, you're like, in five years we already have self driving cars, self driving Ubers in the US, right? In five years Sam will have self driving, flying Ubers that can go underwater type of thing, you know? Or going to space. Just because that's how fast [inaudible 01:02:56] progresses. I think people really need to think in those terms. When legislation lags by two years, or three years, it's actually equivalent to eight years in real life.
Jason Widjaja: Yeah. Five years is a very, very long time in this domain I think.
Kirill Eremenko: Yeah.
Kirill Eremenko: Okay, well Jason, thank you so much, we've run out of time, thank you so much for coming on the show. Really interesting discussion. Where can our listeners catch up with you, get in touch, and follow you if they'd like to see where your career takes you, and what other projects you're going to be working on in the future?
Jason Widjaja: Happy to connect with anyone, connect to, on LinkedIn. And also, I guess the way my mind works, I like to read and write, so I write on Quora, the Q&A website quite frequently. So you want to follow my writing, I'm on Quora as well.
Kirill Eremenko: Yeah, Jason's being very modest. Jason is, as far as I understand, the top contributor on Quora in the space of technology or data or artificial intelligence, what was it, Jason? Tell us.
Jason Widjaja: It was analytics.
Kirill Eremenko: In a space of analytics. So guys, Jason is top contributor for Quora in the space of analytics, definitely follow him there, check out his answers and make sure to connect in LinkedIn. And Jason, for you I have just one question left today. What's a book that you can recommend to our listeners to help them better themselves?
Jason Widjaja: Wow, just one. Let me see, that's a tough one. I think I might go with something that is a bit more general and not so data science-yeah. I think in terms of helping people to digest and consume analytics, if the person you are sharing with is not able to frame what you're presenting them in the right way, then it's not so useful. So I'll probably go with Daniel Kahneman's Thinking Fast and Slow. I think that is a really excellent book in terms of reflecting on the ways we make decisions. And if you're not able to change your mind on the basis of whatever model you are looking at, then you kind of miss the value. So I go with that one.
Kirill Eremenko: Interesting, I've never heard that one before. Is this an analytic specific book or is for anybody who wants to be able to change their minds.
Jason Widjaja: It is a decision specific- I guess it's really much more on the decision science space than about analytics delivery. But yeah, for consumers for analytics, I think that's a pretty good one.
Kirill Eremenko: Awesome, awesome. And you also already mentioned The Lean Startup, so we got two books Jason recommended. So, Lean Startup is for people who want to get into the startup space and Thinking Fast and Slow by Daniel Kammerman, is that right?
Jason Widjaja: Kahneman, yeah.
Kirill Eremenko: Kahneman, Daniel Kahneman for [crosstalk 01:05:48].
Jason Widjaja: So that's k-a-h-n-e-m-a-n.
Kirill Eremenko: Okay, thank you. And I'm sure Thinking Fast and Slow should come up as well.
Kirill Eremenko: Okay, so on that note, thank you again so much, Jason, for coming on the show and I'm sure this is going to be very interesting if not revolutionary information for a lot of people out there. And also I'm sure a lot of our listeners have already encountered you on Quora that just haven't known about you, or now they will encounter you on Quora and they will know who is replying to them. So thank you again for that huge contribution as well.
Jason Widjaja: Thanks very much for having me today, Kirill.
Kirill Eremenko: So there you have it. That was Jason Widjaja, Associate Director of Data Science at Merck and Co. I hope you enjoyed today's chats and got some valuable insights out of this. Whether you are starting into data science, whether you are looking to create a startup in the space of artificial intelligence, whether you're a leader in the space of data science, there was something for you here.
Kirill Eremenko: And personally, my favorite comment from this podcast was about the functional domain knowledges. Very commonly we discuss domain knowledge as the business domain knowledge, as the industry specific domain knowledge that we need to keep up to get onboard and Jason did comment on that that he is looking to get more skilled in the space of pharmaceuticals just to save his colleagues time. But also he mentioned functional domain knowledge and it's very important to keep that in mind as well. That whichever industry you go into, whichever business you go into there are some commonalities, such as marketing, or sales, or HR, and things that help the business run that allow it to exist. And if you get into that space as a data scientist, then those skills are highly transferable. Not only your data science skills are going to be highly transferable, which is always the case, but also your functional domain knowledge will be highly transferable. So that's something to keep in mind for your career as well.
Kirill Eremenko: If you enjoyed today's show make sure to catch up with Jason on LinkedIn, get connected there, follow his career, and yeah, maybe you will get some more additional tips and insights about your own journey into the space of data science, analytics, and artificial intelligence. Also, follow Jason on Quora, he's got thousands of followers there as well. And if you want valuable career advice, as usual you can get the URL to Jason's LinkedIn and any other materials that we mentioned on the podcast at www.superdatascience.com/149. There you can also find the transcript for this episode.
Kirill Eremenko: And if you did enjoy the podcast, then make sure to leave us a review on iTunes, it would be very valuable for us to get your feedback to help spread the word out there and get this podcast visible to other data scientists and analytics professionals. We'd really appreciate if you could spend a few minutes to do that.
Kirill Eremenko: And on that note, I look forward to seeing you back here next time. Until then, happy analyzing.

Kirill Eremenko
Kirill Eremenko

I’m a Data Scientist and Entrepreneur. I also teach Data Science Online and host the SDS podcast where I interview some of the most inspiring Data Scientists from all around the world. I am passionate about bringing Data Science and Analytics to the world!

What are you waiting for?

EMPOWER YOUR CAREER WITH SUPERDATASCIENCE

CLAIM YOUR TRIAL MEMBERSHIP NOW
as seen on: