How to Build a Successful Data Science Team w/ Marcelo Pimenta @ExperianDataLab (Episode 32) #DataTalk

Listen to the podcast:

Every week, we talk about important data and analytics topics with data science leaders from around the world on Facebook Live.  You can subscribe to the DataTalk podcast on iTunes, Google Play, Stitcher, SoundCloud and Spotify.

This data science video and podcast series is part of Experian’s effort to help people understand how data-powered decisions can help organizations develop innovative solutions and drive more business.

To keep up with upcoming events, join our Data Science Community on Facebook or check out the archive of recent data science videos. To suggest future data science topics or guests, please contact Mike Delgado.

In this #DataTalk, we talked with Marcelo Pimenta, Experian’s Head of DataLabs in Latin America, about ways to build a great data science team.

Mike Delgado: Hello, and welcome to Experian’s weekly Data Talk show, where we talk with awesome data science leaders all around the world. Last week, we were talking with Luca in Singapore; today we’re with Marcelo Pimenta, who is in São Paulo, Brazil. He is the head of Experian DataLabs in Latin America. He has his Bachelor of Science in physics. He also has his Master of Business Administration. For those watching live, if you have any questions about how to build a great data science team, which is today’s topic, feel free to push your questions for Marcelo.

We’re going to have a lot of fun in the next 30 minutes. I invite you to ask those questions. For those who are listening to the podcast, Marcelo will be sharing some resources throughout today’s show. If you’d like to find those resources or also find a full transcription of today’s show, you can go to the Experian blog. The short URL is just ex.pn/datalab, and that will be a page where you can learn all about Marcelo and his work.

Marcelo, welcome to Data Talk. It’s an honor to have you. Thank you so much for your time. Can you share your journey, what led you into working in data science and what particular areas you became passionate about?

Marcelo Pimenta: Okay, first of all, Michael, thank you so much for this invitation. It’s a pleasure to be here with you and you guys and try and share a little bit about my passion about science and data science also and tell you a little bit about the story of the data lab in Brazil. Let me give you my background. As you mentioned, I’m a physicist. In fact, before I wanted to be a physicist, I wanted to be an astronaut or science fiction writer. I tried both, but I couldn’t win. I decided to something more basic, like physics.

Mike Delgado: Amazing, because you have those two sides, right? Very mathematic, science-based but also creative and literary. Those are two very different disciplines. That’s fascinating already to hear you have those two passions.

Marcelo Pimenta: Especially science fiction. It’s something people must read more. There are several authors who anticipated the future that we live in. If you imagine that some guy like Philip Dick during the ’30s and ’40s wrote books like Blade Runner, Total Recall, Minority Report, A Scanner Darkly — several books that describe exactly the society we live in now and will live in in the next 20 years. It’s great. It’s a source of inspiration, of course.

When I finished my physics course, I decided to open a company in the ’90s. I started to work with artificial intelligence in the ’90s.

Mike Delgado: Wow. People weren’t really talking about AI in the ’90s.

Marcelo Pimenta: Exactly. It’s not something new. There are several studies from the ’50s to hear, and during the 90’s was kind of an explosion of neural network techniques and neural networks investment. The problem in the ’90s is the cost to store data and the ability to process data. It’s not powerful enough to do a good job. In the ’90s, we thought we would overcome this problem. I opened a company and they tried to figure out the high hand writer symbols in order to recognize values of checks and then reduce the cost of the banks, for instance, to process in a background the information about checks.

We worked four years. I think we did well at this time. We reduced, in this particular project, 20 percent the cost of the bank with it. We decided to expand the business in the beginning of 2000. You know Brazil suffered several crises, especially recession periods. We couldn’t compete with big companies that survived this period. We decided to close the company. I joined the corporate world. I worked at two major Brazilian banks, and then I moved to a big telephone company called Telefonica that most of you would know. I was in charge of the BI, business intelligence, area. Finally, to global innovation, Telefonica.

After nine years, I was a little bit tired and decided to open a new company. At that exact moment Eric Haller extended an invitation to join Experian. When Eric started to describe the mindset of the data lab, it was exactly what I was imagining for my own company, but I decided to go because the mindset was the same. Experian has a fantastic brand. Personally, low risk for me, and the company pays a salary.

I’m here and I can tell you it’s one of the best journeys because we started two years ago from scratch. In two years, we were able to build a huge office here in São Paulo, we developed two products that we moved to Experian’s portfolio. We have been recognized several times as a disrupting team. I think we do a good job.

Mike Delgado: It’s phenomenal. Can you describe the lab you’re in? I’m looking at the background. I see some really cool pictures, and I’ve seen some other photos of the lab. That’s also something that makes your particular program attractive to data scientists, because of the environment you’ve created. Can you describe the lab for those who have never been there?

Marcelo Pimenta: Yeah, I’ll describe it to you and send some links where you can look at it. In fact, I sent you a link with a virtual reality tool so you can go through the lab and see the elements we introduce here. We live in a kind of data science, math, scientific approach, tech guys. We live in a sort of subculture. We have our own signs, symbols, values and so on. We decided to create a place where someone with this kind of background immediately recognizes it. Say, I know these notes, I know this graph, I know these guys. It’s a kind of message. Before we start to talk about business or job opportunities, we like to create a place where people coming immediately recognize it like a home place.

We built this office. It’s about 800 meters squared. We have several areas, we have open space, we have five garages in our land. The idea is that most big startups around the world have started in garages, so we have garages in our project.

Mike Delgado: That is so cool!

Marcelo Pimenta: I sent the pictures and the links, so enjoy. We created some open space, some isolation areas as well. We have a big lab through the garage. This web is kind of a meeting area as well. People need to jump through a jungle.

Mike Delgado: It’s like a jungle gym. You’ve created a very creative workspace in every possible way, from the imagery to having garages and net that people can jump on. You’ve really created a very creative workspace. I think that would be a dream for any data scientist to work at, just because it’s very playful.

Marcelo Pimenta: Some friends of mine here in Brazil say that data labs here are kind of Disney World for data science.

Mike Delgado: Building this data lab from scratch, I’m curious, today’s topic is all about hiring great data scientists. Before we get into academics and skill sets, let’s talk about personality types. When you’re sitting down with somebody, what type of personality are you looking for that would be a great fit on your team?

Marcelo Pimenta: First of all, intellectual curiosity to understand things and to study or really figure out what happens in a particular situation. Intellectual curiosity that people try something different or try to see the problem in different angles or perspectives. People who have these kind of initiatives have an advantage. It’s a kind of requirement for the scientific method. We first try to understand if this particular candidate has this curiosity to understand things. We are just chatty, talk about something this person doesn’t know. If this person asks smart questions, it’s a good signal.

We also look for people who are able to be self-motivated and also people who can work alone, without management. It’s something that once you present the problem, stop, try to understand, and elaborate lists of activities to test and understand the problem. It’s fundamental because sometimes people are facing a business problem and immediately try to work on it. What we are trying to do here is, first of all, figure out how to turn this problem into a math problem. If you are able to understand a business problem and describe this problem as a math problem, we are able to solve it. That’s the way we work on it.

There are some people who use some kind of course like physics, statistics, math, engineering, software engineering, have this kind of training. It’s likely we will find people in these areas. For instance, we have one analyst here who did tourism courses, but she’s very smart. She asks the correct questions, so we brought her to the team and she’s helping a lot. It’s a mix of specific backgrounds that have training to apply and attitude to face the problems. It’s important. It’s one of the aspects.

Regarding building this from scratch, one point we are working on is that Experian has a kind of corporate feeling. Especially for the young guys, when they see Facebook, Amazon or Google opportunities, they look and say, “Hey, it’s not too fancy.” One of the things we are doing is breaking this barrier. Showing people that it’s different and that in fact we are doing good things here.

One of the requirements to be a SETC company is to have data, and we have the biggest data sets around the world. If you are able to work with data, it’s one of the best places to work. We start teaching people from university, even high school, that it’s a good place to work in terms of career, challenges and to experience something new. We have some programs for high school students and summer programs for graduate students where we give real problems to them and they solve them.

I brought one example for you. Let me share. Are you seeing this?
Mike Delgado: Yeah, that’s cool. What is that?

Marcelo Pimenta: It’s a kind of autonomous car. Part of its function is to figure out the intensity of the Wi-Fi map in any place. It’s autonomous because it’s able to walk around a place without running into any wall or chair. It’s one example that a high school student did. The best part of this project is that we start the project showing a particular scene from a Batman film, The Dark Knight, where Batman tracks through Gotham and creates a 3D image of the buildings, the cities, the streets and so on.

We show this to the students and say we need to build a car that would be able to walk through a place, collect the strength of the signal of the Wi-Fi and then build a 3D image of the place. We are talking about guys from 15 to 17 years old. They did it.

They got back to the schools and said, “It’s a great place to work. We’re doing something really cool from Batman and The Dark Knight.” It’s a real business problem. Nowadays we’re using some tools to figure out some malls and remote shops, and then we can create augmented reality interaction. We thought to send someone to the place and collect all the information of the place because we are using an autonomous car to do it. It can change the minds and perspectives of the people.

Mike Delgado: This is amazing, Marcelo. And for the people listening to the podcast, you aren’t able to see that what he showed on screen was a remote-control toy car, but it’s so techy. It has these different sensors on it; it has cameras because basically it’s a little autonomous vehicle. What I love about what Marcelo did, going to the high schools, inspiring these kids to get involved in these data science projects using things like Batman, things that they’re maybe watching on TV, as a springboard to inspire them. You’re sparking this intellectual curiosity early. Marcelo, can you also talk about these hackathons you host? I saw a video where you had hundreds if not thousands of people in your labs.

Marcelo Pimenta: I think one of the missions of the data lab is our work on the cultural change of the employees and the supplies, and they are the people who are connected with us. We started supporting the open invitation and festivals, so there are several events in the same place at the same time.

One of them is the hackathon. We try to solve real-world problems. We invite people to solve three or four challenges simultaneously, and we promote them and bring them to Experian. We teach them and prepare them in order to understand the business proposition of the problem. They spend five to six hours to build a dashboard, and then they start to code. At the end of 36 hours, we evaluate the code. They must be the prototype. We evaluate in terms of if it has met or not the challenge and how good the software is.

During these hackathons, we figure out leads to contract to our teams. We also find some small startups that are trying to show them to us to be a kind of supply. We check and evaluate them and invite customers to understand what we are doing here and the ability to produce innovation.

Simultaneously we give a class. We give a code class for the general public. We promote some workshops to explain technology in the commerce and the business, in the life of the people. We also have a demo and pitch day where we receive some companies that throw a pitch to us. Everything in the same weekend.

Mike Delgado: That is so cool. What I love about you, Marcelo, is that you’re not only a team leader, but you’re so motivational and inspiring to be able to create an event like this. Folks who are listening, just imagine hundreds of people all around, kids different ages, teenagers, dedicating 36 hours. They’re gonna be in a room for 36 hours, working on a business problem, with different data scientists, helping to solve things. It’s so cool what you’ve done, Marcelo, and it shows your strength as a leader.

I know we’ve been veering off course of the topic. I think this all speaks to if someone is choosing to work for a company, they want to choose a company that is innovative, but they’re gonna work for somebody that’s gonna push them and challenge them and inspire them, and that’s what you’re doing every day at DataLabs — the hackathons and all the other things that you’re doing. Obviously, this playful environment that you created. It’s unique for me to talk to someone like you who has both that mathematical science mind but also that very creative mind. Often when you go to school you have these two tracks. Do I go into the arts? Or do I go into the sciences? And you’ve definitely merged them together and made a beautiful space.

Getting back to the topic about hiring, you talked about things like intellectual curiosity, being self-motivated, someone who wants to be part of a team, someone who is naturally curious and not gonna settle, someone who is gritty. Can you also talk about academic backgrounds that stand out to you? And if you were to describe the perfect team of academic backgrounds, what would you be looking for?

Marcelo Pimenta: I think that all the guys here must meet minimal requirements. These minimal requirements, it’s some training in the scientific method. It’s the way you face the problem and decide. It’s the way you reduce the problem and build a simple model to reproduce this problem. Once you do it, you start to do the hypothesis and test this hypothesis in the single model. People who do engineering, math, physics, statistics have this type of training. That’s the basics.

Particularly here in the group, we work on complex networks. Artificial intelligence and machine learning, and we are also interested in information theory. Especially in the society where we live with huge data sets. Every single data set around the world has problems with missing data or interpreting the data. So we apply information and game theory in order to rebuild missing data sets. It’s a little complicated, but it’s very important to the society that we are living in.

We are looking for guys with this kind of experience. We have some projects we use more of a particular kind of skills, but the basic is what I mentioned to you.

In terms of graduation, usually people who have a Ph.D. have these abilities that I mentioned, but it’s not necessarily the criteria that you apply here. I can give you an example. We had two guys here who are pretty great. In fact, we say they come from another planet, no? They just have golden medals in math athletics and so on. They just finished their undergraduate course last year, but they’re producing skills at the same level of Ph.D. guys. That’s the kind of people we’re looking for. Someone who provokes and questions the rules. Someone who breaks the rules and finds breakthroughs in the problems we are solving.

Mike Delgado: That’s amazing. I didn’t even know that there were medals for math. I was a liberal arts student, I was all English lit so …

Marcelo Pimenta: Michael, last week I interviewed a guy and he’s in the second year of university. The resume was name, references and objective, and several medals. Golden medal, golden medal. Okay! Bring this guy. He said, “I can just spend 16 hours a week because I am taking class.” I said, “No problem. We have space for you.”

Mike Delgado: You’re like, “Come in. Just come in, please.” Amazing. We didn’t get medals for reading books in my major. Mathematical medals, amazing. You mentioned the resume. I’m really curious because I’m sure you look at tons of resumes all the time. What are some things that — because there’s a lot of aspiring data scientists in our community who are always wondering how to stand out. To a hiring manager, to a director. What should I put on my resume? What are some things that stand out to you on a resume that make you go, “Hm, I want to talk to this person.”?

Marcelo Pimenta: Twenty golden medals is a good way. Usually a background is important; the first university is important. When you see 10 to 15 universities, you see someone come from this place, you say, “Let me take some time to look at this resume.” This kind of background is important for data science. When you talk about software engineers, the most important thing is the code. I evaluate the person based on the code that person writes. No one is equal. For data science, the first university is important; it’s a kind of business card. I did Yale, Stanford, Harvard, MIT, Duke — or here in Brazil, São Paulo University, San Carlos — or things like that because we know someone who finished this type of course has the ability to work in a place like ours.

The second point, I try to pick the experience of this guy in terms of solving real problems because sometimes data science is too academic. Of course, we need the fundamental in academics, but we need to solve real problems. I try to check how many real projects this person did. The role of this person in the project — usually I ask them the three biggest problems they face in their life. How they solved this problem and what the strategy was to solve this problem. That’s pretty simple. We open a position, we receive around 1,000 resumes. The HR team does the first selection. When I get back to them I say, “Write your three biggest challenges that you’ve overcome in your life. Tell me about the problem, how you solved it and your strategy you used to solve it.”

Just three questions and around 10 to 15 percent of the people come back with their experience. It’s something curious because so many people say data science, for example, but they can’t spend 20 minutes, 1 hour, to explain what they did and how they solved the problem. You start to figure out who is serious or not with attitudes like that, no? Once we have five to 10 candidates, we start interviews to pick up some behaviors that I mentioned to you, especially this curiosity.

Mike Delgado: I know we’re almost out of time here. You mentioned one of the first things you’re looking for is schooling, and obviously going to top schools like the MITs, Caltech, the schools you mentioned in Brazil, outstanding schools, if you’re getting into those schools you’re curious, driven, self-motivated, etc. But what about those who weren’t able to get in to those top-tier schools? What advice do you have for those people who are trying to break into data science but just don’t have pedigree?

Marcelo Pimenta: As I mentioned, sometimes 20 golden medals is good. I can give some advice. If you are good to code, that is a good way to get some experience in some difficult problems without necessarily having data science skills. If you’re coding for the data science team, it’s good. You start to think like them, you learn like them, you learn how they are working and how they produce some ideas. Try to build a nice high storage track in terms of problems that you solve. Nowadays, in fact, for someone who had no experience at university grade, I usually say, “If you conclude completely one of the formation from Coursera, I’ll talk to you because despite the fact that it’s an online course, I did some Coursera course.” It’s not easy to conclude the entire path. If someone says, “I completely concluded six, seven courses in Coursera regarding data science, I’m seven in code.” I say, “I’ll talk to you” because it’s one of the ways to get my time to this person.

Mike Delgado: It shows a lot of motivation and grit to stick through a Coursera course.

Marcelo Pimenta: Yeah, another thing, you need some kind of discipline because usually you are studying or working and we are taking your free time to apply and understand something new and apply your time to develop new skills. Someone who has this kind of disposition is someone I try to talk to and hear about his or her story.

Mike Delgado: I love it. We have to close. This has been an amazing episode with Marcelo Pimenta. Again, he is the head of the DataLabs in São Paulo, Brazil. Just some things that stood out to me in this conversation as he has built a great data science team. If you’re looking to break into data science, Marcelo pointed out intellectual curiosity, being self-motivated, being a team player, being gritty, jumping into projects like Coursera, and just putting your all in and showing you can complete the work. Just hustling and showing you can complete the work. Getting involved in hackathons and putting in the work. That’s gonna make you stand out as someone who is going above and beyond just “I got the degree” or “I have the skills.” “I’m passionate enough to dedicate my time to a 36-hour hackathon.” That shows grit, that you’re self-motivated and you’re gonna be part of a driven team because that’s what Marcelo is looking for.

Marcelo Pimenta: That’s it, Mike.

Mike Delgado: Wonderful. Well, thank you so much again, Marcelo. I want to let everyone know that if you’d like to learn more about his work, you can always go to the transcript of today’s podcast, which will have resources and links to different pictures of his awesome data lab. The short URL is just ex.pn/datalab, and then if you’re interested in joining our data science community on Facebook, the short URL is just ex.pn/datagroup. That’ll bring you over to our Facebook group where people are asking questions. We have a lot of people who are just getting into data science, and we invite you to join us there.

Marcelo, thank you so much for your time today, and we hope to have you on the show again soon.

Marcelo Pimenta: Thank you so much, Michael.

Mike Delgado: Okay. Take care.

Marcelo Pimenta: Take care. Bye-bye.

To suggest future data science topics or guests, please contact Mike Delgado.

About Marcelo Pimenta

Marcelo Pimenta serves as Experian’s Head of DataLab in Latin America. He earned his Bachelor of Science degree in Physics from Universidade de São Paulo and his M.B.A. from Fundação Getulio Vargas.

Check out our upcoming live video big data chats.