Loris Marini_mixdown.mp3

Loris: [00:00:00] We all matter, we are all part of this and we need one another. I can't do that a science well, if the data is not reliable, if it's not trusted, if it's not connected to the business via metadata by a data management program that touches anyone. And so it's really a mind, a change of of mindset from. You can add value as a team in isolation to, well, not really. Data is the common denominator to everything we do, whether we like it or not. Everything we do generates Data.

Harpreet: [00:00:46] What's up, everybody, welcome to the Artists Data Science podcast, the only self-development podcast for Data scientists. You're going to learn from and be inspired by the people ideas and conversations that'll encourage creativity and innovation in yourself so that you can do the same for others. I also host open office hours you can register to attend by going to Bitly.com/adsoh forward slash a d s o h i. Look forward to seeing you all there. Let's ride this beat out into another awesome episode, and don't forget to subscribe to the show and leave a five star review. Our guest today has been passionate about information and data for over 15 years, and he's on a mission to make data reliable, discoverable and ready to power analytics. He's earned a PhD in physics with a specialization in quantum photonics and was hired as the first data scientist at an organization before making his way into Data architecture. As soon as he entered the Data world, he learned firsthand [00:02:00] that the data science hype that gave him a job also created a gap between perception and reality. Having lived these problems firsthand, he decided to start what is now the Data Project, a podcast to bridge the gap between Data users and Data experts. Through the Data project, he speaks about the non-technical challenges in data and information management, hoping to inspire a healthier and more effective way to work with data. So please help me in welcoming our guests today a man who views data through a human lens. Loris Marini Loris, thank you so much for taking time at your schedule to be on the show today. I appreciate

Loris: [00:02:40] It. It's a it's a pleasure to be here.

Harpreet: [00:02:43] Yeah, man, I'm actually excited to chat with you. You know, I'm going through a lot of the same challenges that you have gone through before as the first data scientist in our organization and then bridging that gap. So I'm excited to chat with you. But before we get into all of that man, let's learn a little bit more about you. Where did you grow up and what was it like there?

Loris: [00:03:04] Yeah. So I I grew up in in Rome, I would say, you know, technically not the center. So we're talking about forty five to an hour drive away from the center on a Sunday in the middle of August and then Monday on a typical Monday and protect two hours because it's congestion. And so it's a big mess. Now Rome is beautiful, man. It's it's a beautiful city, obviously. I mean, it's very famous in the world for for for the the history and the museums and the cathedrals and the arts in general. But it's also a pretty, pretty intense place to live because, you know, Italy is not famous for its smooth bureaucracy, I guess. So we have our challenges and and that means that basically everything you want to do takes a lot of time and patience. So it's never a straight line from A to B, you're going to have to you have to go around many, many corners. Sometimes new corners appear, the landscape changes in front of your very eyes [00:04:00] and you're like other how's this making any sense? But yeah, so this is the context. Forty five. So it's a city that's actually, to my surprise, very famous. It's it's called Tivoli, and it's a ten minutes drive from Villa Rihanna, one of the ancient pillars of the Roman Empire. Adriano, the emperor, used to used to live there and I literally like my my family house is like five minutes walk from the ruins of this, this old emperor. The place is beautiful, but you would expect something, you know, the walking and Villa Adriana and find, you know, something that resembles a villa. But it's actually just that big piece of land because it's been so much from time, I guess took its toll, and now they're really just a bunch of ruins.

Harpreet: [00:04:49] I mean, Italy is beautiful. I've been to Rome once, many years ago. It's probably like 2008. That's the first and only time I went to Rome. I did drop some coins in the Trevi Fountain, so I guess that, yeah, I'm due back at some point in life. But yeah, like, I love Rome. I love I love the Mediterranean in general is that that region is absolutely beautiful. I love it there.

Loris: [00:05:11] So, yeah,

Harpreet: [00:05:12] So what the heck is quantum photonics and how did you get into that?

Loris: [00:05:17] It's a good question. Quantum photonics is the study of light and how light interacts with matter, and because it's at the very the small scale, because we're interested in very feeble light sources and very, very tiny pieces of matter. It gains the word quantum, meaning that it just follows a whole bunch of very unintuitive properties that you wouldn't find in the macroscopic world. The big why people bother to study quantum photonics is that in the last 20, 30 years, we have enjoyed a very smooth and predictable Maurice [00:06:00] Low, which, as you know, predicts a doubling of the number of transistors in any CPU every 18 months. And it's being, you know, holding true, very empirical. There is no particular close mathematical formula for that, but you know, more notice this pattern and has been going for a long time, and it's only in the last five to 10 years we noticed a saturation of that trend. And and now it's a reason why our MacBook Pro 16 inch today doesn't have, you know. Ten times a 12 times the number of transistors they used to be used to have 10 years ago, five years ago. And so what kind of people are wondering, you know what, what's going to be next? You know, we can't pack more transistors in the same chip. What else is there? And luckily for us, in the sixties and seventies, the scientific community is particularly the theoretician developed a framework to try to understand quantum mechanics, the behavior of electrons or protons or photons of phonons.

Loris: [00:07:00] So these are all very scientific names to mean very, very tiny perturbations like a phonon is the is a perturbation in vibration. So there's more or less amount of vibration you can think of below, which there is nothing that's called the phonon. And a photon, obviously is the equivalent for light. It's the smallest electromagnetic perturbation that you can think of. The smallest amount of electromagnetic energy at the same as the electron is is more or less like elements in terms of positive and negative charges. And so all this more or less building blocks, they come together and they exhibit very interesting properties. For example, you know, a famous one is, is teleportation or another big one is the spooky action at a distance that February. There's a very famous paper of Einstein that was published a few years before he died. He was he was trying to understand how the heck is possible that you take two particles, you make them interact [00:08:00] in some way and then you bring them at an infinite distance, one from the other. And there are certain conditions under which you can. If you change the state of one, say you flip it or you mirror the other, one does the opposite and complement the complementary flip by itself, no matter how far they are. And so he couldn't understand this. And so, yeah, it's mind blowing

Harpreet: [00:08:25] And that crazy. It gets weird. You start getting at that really, really small, small level like that that I was interviewing Max Frenzel, who's also a physicist. He's currently based out of Japan, and he researched like quantum

Loris: [00:08:42] Promote something dynamics.

Harpreet: [00:08:44] Yeah, yeah. He was looking at how to build the smallest engine possible, and I was like, Dude, this stuff is so cool. So I mean, how did you go from like, like this awesome, crazy physics stuff into Data science?

Loris: [00:08:57] Yeah, it just happened by chance. Like most things in life, right? As much as we like to plan, there's a whole bunch of quantum randomness. I was having a beer with a friend towards the year three of my PhD, and I was in the middle of writing my thesis. Basically, I was studying writing my thesis more, and I had an interest in machine learning because before my PhD, I did some research here in a union with the group of Branka in telecom electrical engineering. They call it here, but it was really information engineering. They will look. I was looking at applying reinforcement learning to the problem of minimizing latency in and in a cellular network. So you got a bunch of phones and some popular content. You can imagine a library of a hundred very popular files. And the question is, can we cache those files and how should we cache it? Which files should go where so that the overall network experience is the fastest service [00:10:00] possible? And it turns out to be one of those the mathematicians called not pulling on me on hard problems, which means that you can't really, yeah, you can't solve it. With the brute force approach, you have to come up with a better with another way.

Loris: [00:10:14] So the optimal solution is not is out of reach, and all you can do is hope to find the sub the close, the suboptimal to the optimal in a reasonable amount of time. So it was interesting. Like I basically looked at the literature and I found this paper that was suggesting to approach the problem as a game theory problem and deploy a bunch of individual actors or players. They call it learning automata. Terrible name very unmarketable. But the idea is that you got these nodes and they take actions and look at the environment and based on what they see and what they feel they make them, they make a choice to say, I'm going to cache these files. And then as everybody does this and everybody listens to everyone else. If the system agrees on rewards and penalties, the idea is that the whole latency will go down. And so you reach us, something that's very close to the optimal in a super short amount of time by distributing sort of the responsibility of finding the best combination to a whole bunch of individuals.

Harpreet: [00:11:18] That's super, super fascinating. That's wow.

Loris: [00:11:22] Yeah, I really loved that piece of research, but it was very short. It was like six months, and I was mostly coding at the time in MATLAB and. Proving that, you know, I worked in that context, too, as long as we made a whole bunch of assumptions and that was it was useful, interesting. But it connects to my PhD because I was technically a staff member. I was a research assistant and I had access to a whole bunch of courses that the university provides to staff. And one of those was how to be like a path to become a better lecturer. And I always loved teaching, so it's like, I'm going to check it out. So I go into this big room and I find [00:12:00] and I find this what then became my supervisor. I was there. He was new. He was setting up a team and he asked the question to the professor. And I kind of picked up his accent. I was like, This guy, your turn. You're going to check me and say, hi, say hi. And he said, I've got a laser lab. Then I'm putting together that you want to check it out. I was like, What are you talking about? And so that was love at first sight. When I when I saw the the lab, you know, the I just I just fell in love. I couldn't resist.

Harpreet: [00:12:27] And then so that was kind of your entry into just going from from hardcore physics and research meant to Data science and doing Data science type of work. And then eventually you were hired as the first Data scientist at an organization. So talk to us about that experience and how did that experience, I guess, lead you into the Data project?

Loris: [00:12:49] Yeah. So it was a very direct entry line. Like there wasn't there wasn't a position that the company did not realize for scientists. They probably didn't even know that they needed somebody to look at the data and do something with it. It was. It was this friend of mine back to the famous beer I mentioned at the beginning. We were having this chat about my PhD and this this problem of reinforcement learning, and you got super interesting. He was at the time, the lead developer was working on an internal project to to improve the way the data was stored. So it was a distributed system kind of work. Very interesting. So when I say distributed like, I guess that was the key word that resonated with him and said we should talk. So I invited him over for a beer before I knew I had a conversation with CEO and the CTO, and they were like, Sounds like we should give this a go. But it was very unstructured. So I was brought in, as you know, you're going to help us figure out how we can, what can we monetize, you know, how can we extract value from our data? But there was absolutely no plan whatsoever. There was no inventory. There was no. Yeah, it was. It was very random. I'm sure that many people resonate with that [00:14:00] situation and it was also the twenty seventeen, I guess, the peak of the hype list, maybe not the peak we were coming down, but still strong enough here in Sydney or in Australia. You know, strong enough that when the CEO looked at my what I've done in the past and looked at what they wanted to do in the future. So an opportunity I said you should, you should come in. So for that, I'm grateful because it was a fantastic experience. I learned a lot, but it didn't come without challenges.

Harpreet: [00:14:30] Yeah. So let's talk about some of those challenges, man, because that's that's the interesting stuff, right? Because I mean, like, I was kind of hired on the same premise and I'm, you know, still absolutely loved my company left my job. But it was the same kind of thing that you're talking about like, Oh, we've got a possible idea to do something and then you do that. And then all of a sudden of looking around at the rest of the data they have in the organization, you're like, Oh my God, this is everywhere. I don't know how to do anything with this. So, so when you you talk about that, the gaps between reality and expectation, like what did that look like when you were, you know, venturing out as the first data scientist?

Loris: [00:15:04] Yeah. So I think there's many gaps, actually, I would decompose it a listing in two parts. The first is a high level, so, you know, trying to understand what the company really wants to do and map that to the data that they already have. And then, you know, venture into the process of, OK, now we know now that we know exactly what we want to do. Say you want to build the recommendation system, then you start looking at the problem with that with that peer. You know, with those lenses, you know, I need I need to find anomalies so you might look into a large database. In that case, we were using the ElasticSearch and trying to understand how that correlates with the actual app. This case was a MarTech company, so a really a SaaS company so that we had I had access to a lot of data. So that volume wasn't a problem that were right, that it was there. What [00:16:00] wasn't there was the connection between the data and the user. You know how when you when you drag and drop a shape, you know, we had this concept. Still, Autopilot still has a bit of a canvas that you can bring shapes and build a graph, really a piece of logic for your marketing campaign. So be that, you know, trigger an email or react to when somebody fills in a form on a website.

Loris: [00:16:28] Very interesting stuff. But. And it all made sense in the heads of the developers of the of the designers, people that were deep into into the product, but when you come in from the outside and you're looking at this picture, the first question is once you understand the strategy, you know, that's what you want to do when to build a recommendation engine. The immediate next step is OK, what? How does that? Where is the Data? And then once you find it, how does that Data connect to the actual app to that drag and drop experience, to the state machine, really, that it surely is there, but it's rarely documented, particularly in startups. So there was a lot of context building a lot of dumb questions that I had to ask two people that looked at me and said, This is obvious, you know, why don't you just read this repository? So sometimes I would get back a link to a GitHub repo or an issue that was open where somebody detailed the logic of of that portion of the system starting from an issue that somebody else had so long scrolls. And it's really it's really time consuming when you don't have an inventory, you know, some a curator, almost a trustee of of Data, right? Somebody that knows where things sit and how they connect together and can connect the business to the Data.

Harpreet: [00:17:49] And so kind of in that environment in that doing that work, you somehow managed to find your way into Data architecture. So we'll talk about that transition. [00:18:00] What was that transition like? What made you be like, Oh my God, I need to. I need to put the Data science down and pick up the Data architect stuff.

Loris: [00:18:06] Yeah, it's a fantastic question. It was just pure need really like we were when when those first problems are solved, you typically find yourself in front of a, you know, an hour notebook or Jupiter notebook. I was working Python and Time and I had a sample of the dataset. I had my model and I had an output, and I remember three months in, we had the proof of concept. You know, we we have we had the system. It was working. It was it was giving us the type of predictions that we wanted. It was easy to understand. So that interpretation box was was tweaked and the CPO was happy. The CTO ecstatic, like, we've got it. We've got, you know, now we can go and and change the world and what what was missing was the deployment and data ops side. So how do you go from? I think that works in Jupiter to I think that can serve 20000 paying customers with minimal latency. So all that engineering bit. And I think we at the time, we we underestimated it. And so that's where the second element, which is so strategy was the first. The second one is literacy. There's a tendency I find to for people that are really deep in software development to assume that Data is just like software development, but you just need a bigger hard drive. Right? And that's all. That's all that changes. And I think this is wrong and it's dangerous. It's very hard to change that mentality because I see why you would make that the logical connection.

Loris: [00:19:39] But Data is not software like software development, and there are intrinsic challenges. And one one of the immediate things that pop up is the monitoring that you have to do to to guarantee the certain performance for the algorithm because everything, everything shifts, you know, the input data moves all the time. And if the model doesn't adapt to those [00:20:00] changes, you can get wrong predictions or behavior that is suboptimal and you might not know that, but eventually the users will feel it. And what's worse of a messy system is an invisible and discoverable mess. So it's all nice and good to be agile and go for the for the low hanging fruit. That's very important in any change management project. And this was a change in management, really, because it's a new function for the business. It's a new it's a new way of thinking, but you've got to think big picture as well. So that's where the architecture came in was a need to say we should take a step back and now we know we've got something valuable. Let's see what is the best way we can deploy. What is the infrastructure that we need to make sure that we're not going to be awake at three a.m. responding to a kingdom notification because a server is down or because the insights are not there, the cache didn't load properly or didn't refresh properly. And so all of this very practical engineering concerns.

Harpreet: [00:21:06] So speaking of engineering, dumb question here, because even I'm a bit shaky on the definitions here, like what is the difference between Data engineer and the data architect?

Loris: [00:21:16] Yeah, I don't think there is a universally accepted definition of the tool, but the way that I think it is an architect should really be focused on the fundamental foundational kind of choices that are hard to change, that are really expensive to change. Like an example that I make often is the high way. And if you're building a new highway to connect to cities, you will have to make a bunch of decisions like, you know, how many lanes, how wide should the lane be if you're designing an exit? What's the curve radius of the exit? Are you going to, you know, turn 90 degrees in 100 meters or in a kilometer? Those are foundational questions. One, Once you decide that [00:22:00] the skeleton, you can just change it quickly. You can just make an exit to 200 meters long. Instead of a kilometer long. You have to redesign the road, you have to rebuild the systems around it. So changes are really expensive. Engineering is more concerned about how once you know what the skeleton looks like, how do you make? How do you optimize it? How do you make sure that the systems run smoothly, reliably? And so. Yeah, that's that's why I'm curious about your definition. How do you see things like how?

Harpreet: [00:22:31] I really don't really truly understand the difference between a Data architect and a Data engineer. I know that I just say okay when I hear Data architect to me, like it equates to like software architect, meaning that this is the person who kind of like draws the maps and, you know, does the. Diagrams and kind of makes the plan for everything, but then leaves the actual implementation of those details to somebody to code up and that person who codes that up and maybe a Data engineer, I mean, not, I don't know.

Loris: [00:23:00] Yeah, I think I would agree with that, except that I'm more and more I'm thinking, does the obstruction of detail have to be there to make the distinction? Because a famous example is dev ops, the movement that we are all familiar with in software development. The concept that you don't you can't separate development from operations because development is the phase where you're thinking about a new idea. You have the high level, you know, concept operations is the team that actually has to make that happen reliably at scale. And so if you separate those two. What tends to happen is that you make assumptions in development that are not going to be true in operations, and people in ops will have problems. And this is true in software, but it's true in in food as well. There's a very practical example from from my wife. She used to work at a food company and in the NPD, the new product development team. So all about them developing [00:24:00] the concept, developing the the new idea, the new recipe and then ops was in charge of, OK, now you have to make, you know, 1000 liters of this soup by next week. And so they had that kind of problem. So I think it's more of a it's across industries. So we need to bridge this too. And with going back to the architecture that you just mentioned, I think in architect that just focuses on the high level stuff without taking into account the details is not doing a good job. I think, yeah, I you're going to be able to go up and it's just no matter what you focus more.

Harpreet: [00:24:34] Yeah, you got to have kind of both the clouds and the dirt there. I mean, obviously, like I, I definitely could see the ignorance I have towards what these roles do. What do you think a data scientist at a minimum, I guess should know about Data architecture and the role that Data architect plays?

Loris: [00:24:56] It's a good question. I think it's hard to answer because the the the word that a scientist is is very cloudy as well, but I'm going to make a bunch of assumptions to simplify the problem. So if we assume. That a Data scientist, somebody tasked with taking some road Data and extracting useful information out of it, useful and actionable information. And that's it. You know, just just just that bit. It's not responsible for putting the stuff in production. It's not responsible for making systems reliable. So it's really just the scientist, not the engineer. Then I would say knowing about modeling is enough, but this is an obstruction is the assumption I just made is very far from the truth in in real life, you cannot decouple those two. And so in a sense, I'm not surprised that the term data scientist is not well defined because it kind of appeals to the the end goal. I want [00:26:00] to extract Data information from that. Ok, great. That's the useful bit. But how are you going to do it? And you don't need just science. You don't need just engineering or architecture or design. You kind of need everything at once. And so I'm a big fan of this holistic view. I don't think anyone can. It's not useful to put people inside a box and say, this is exactly what you're going to be concerned with and everything else. Just ignore it. It's it can be useful as long as we we make those edges a little bit more soft, you know? So we have this is your concern. But as a data scientist, you're concerned about modeling, maybe, but you should take into account what's around you. And I don't know what. What's your view?

Harpreet: [00:26:49] Blurry to me? I think I forgot what my question was somewhere in there, but I don't know. Like when. So we just hired a data architect at my company is a software developer that we kind of moved up the ranks and he's acting as a data architect, and he's pretty much so has like the software development mindset and me as somebody who's just a statistician, academic type of statistician, having to understand where software development is coming from and then now trying to understand where a data architect is coming from. It's it's been extremely challenging for me. But yeah, to answer your question about what I think a data scientist should know about data architecture, first of all, just. Maybe trying to understand what it is that an actual Data architect does and how the work that they do enables your work to be a bit easier, right? Because I think as data scientist at the end of the day, we ultimately are and users of Data, right? I mean, I'm beginning to realize that more and more now as I try to help my company develop a Data strategy is that all my life, I've just been somebody who uses Data as an end result, right? Like, it [00:28:00] has some origination and lineage far before I come in contact with it. And I think that the way I see it is that maybe Data architect is somebody who who is more involved in that. You know how the real world phenomenon is generating Data and how are they capturing that and storing it and disseminating it through an organization? That's kind of my understanding of what the Data architect does, but I definitely could be completely wrong.

Loris: [00:28:29] Yeah, it's definitely one way to look at it. It's also easy to absorb. The one that I have in mind, I sometimes I get a lot of pushback when I talk about my interpretation of that architecture from from engineers because we like to put labels on ourselves and to identify what we do and who we are and ourselves, what we do. And so the question is, I am an engineer or I am a scientist, but I think those labels on less and less useful when you try to focus on the outcome, we all really all in this together. I want to want to use Data to allow somebody to make more informed decision at some point, whether it's the paying customer or somebody internally. So if that's the goal, there isn't really it's not helpful to put ourselves in boxes. I guess that's what I'm trying to say.

Harpreet: [00:29:24] Yeah. And that are the soft edges that you're talking about, and that really just talks about your human centric kind of approach to Data that that like that.

Loris: [00:29:33] Yeah, I think it's necessary when you do need to be flexible and take different learn to take different perspectives. Doesn't mean that we have to one person has to do everything because it's unrealistic. That's not what I'm saying, but I'm saying that it's useful to learn our brains to absorb and actually wear a different perspective and different view. Because there are it can [00:30:00] be eye opening, can be very insightful to start thinking like an architect, even if you are just in the visualization team, right? They ask questions that may be helpful for you, too. So thinking about the connections, the dependency between systems, you know, if I make a change here. How many systems will be impacted, that impact analysis is something that's useful, whether you are putting together a dashboard or you're designing a data warehouse or a Data ETL system. I think it just applies across across the pipeline.

Harpreet: [00:30:38] So talk to us a bit about that. The Data project, the podcast and kind of like the initiative that you're starting with, the Data project. Is it in an attempt to kind of help soften these edges that you're talking about and help blur the lines between these different identities that we box ourselves into and just help us all realize that we're here at the end of the day to make someone's life easier with Data. Talk to us a little bit about that. If I got your mission like right or

Loris: [00:31:03] It's the it's it's the perfect description.

Loris: [00:31:07] I don't think I can describe it better than you just did. Yes, it's exactly that. It's the realization that we need. We need if we it's an assumption. It's an untested assumption that a project is an experiment, but it comes from the idea that if we know what it feels like to be the other box we can, we can make choices in our day to day that will improve the overall fluency of the system so that we can go literally from the row long in the database to an impact to an actual

Loris: [00:31:43] Human being,

Loris: [00:31:45] Not only faster, but in a way that's more sustainable. That makes us happy to good work, that makes us proud of what we do and feel part of a bigger system that we don't have to fight with the architect. We don't have to fight with Data owner because [00:32:00] the data is mine and it's not yours and you don't have visible, you know, there are challenges in any organization. But I dream of a world where no matter who you are in, doesn't know where you are in the pipeline, you can add value quickly because you understand your place in the bigger, bigger system.

Harpreet: [00:32:16] Yeah, I like that, I mean, just you think of yourself as part of the system, not just separate from the actual system itself, kind of.

Loris: [00:32:24] One note that I that I like to add here is that when I started, I was hired as a data scientist, I suppose, but those are really a huge big difference in average salary between an analyst and a scientist. I don't think that's helpful, to be honest. I might. Yeah. No, no. It really isn't because it reinforces this idea that you are up. You know that you have a higher status as a scientist. Maybe you have a PhD. Maybe you've done some science with some research. And so that makes you as a modern person. I hate that I'm trying to really decompose that that message just saying, Look, we all matter. We are all part of this and we need one another. I can't do that science well, if the data is not reliable, if it's not trusted, if it's not connected to the business via, you know, metadata by data management program that touches anyone. And so it's really a mind, a change of of mindset from you can do you can add value as a team. In isolation to well, not really Data is the common denominator to everything we do, whether we like it or not, everything we do generates Data. And when you touch Data, you touch inevitably finance marketing operations. So we need to think more globally.

Harpreet: [00:33:49] One hundred percent agree with that man. I don't like this. Like, OK, so I Data science. In general, the field is a broad umbrella category. There is a lot that goes under this umbrella of Data science. Like, [00:34:00] Yes, there's the classical title Data scientist, but the data analyst is also in data science. Business intelligence person is also in data science. Yes, data engineers. Also in data science. Data Architect is in data science like to. I don't know what this why it has become a status hierarchy game and why they have turned this data science profession into this. This like just status signaling with the title. And people are like, Oh, I'm just the data analyst. I know I'm just the data analyst. They can't do, you know, like nah, man like people who I feel like people who are constantly like making these type of infographics about data scientists versus data analysts or whatever, they're creating divisions. And maybe they're just trying to status signal themselves like, Oh, I'm a scientist, and it's like, Dude, all right. Well, data scientist means a lot of things. Things like you're just talking about, right? Like, I'm a data scientist. Yeah, but I'm primarily a statistician. Like, first and foremost, like you talk about deep learning the universe, I don't know how to do any of that shit, right? So yeah, exactly. It's just it turns into a status game, which I fundamentally do not like playing those type of games.

Loris: [00:35:12] It's not useful. Yeah, yeah.

Harpreet: [00:35:13] So I got a chance to read it. A lot of the blog post that you have on the Data project, and I feel like there's so many good bits of wisdom and insight in there. I'd love to to get a little bit into that. There's a post there. You're talking about the difference between data, information, knowledge and strategy. Talk to us about that. What's the difference between these? How how do how does data, information or knowledge play into a strategy?

Loris: [00:35:39] Yeah, I mean, this is a when I wrote it a year and a half ago, I thought there was something, you know, kind of new to it. But as I started reading around blogs, I realized that it's many. Many people have expressed very similar ideas. And I guess that's just the reality of everything we do. [00:36:00] But they're really the idea is that that data is useless when you don't, when you don't act on it. That's that's the premise. And acting the the acting procedure really is not the right word, because you can always act on a piece of Data, you know, you look at, you take any dataset, you make a simple aggregation like count by a categorical variable and you've got any insight already that tells you the distribution of a number of records per as Q or an item called price tag or whatever. So you can you can always do something with it, but that's not what we want to do. We don't just want to act on it. We want to act in the best possible way and best possible way means to align our action, our next steps to what the business wants to achieve ultimately. So you can't really decouple anything. We're doing Data from the business. And so going back to that separation, the seeing Data is a specialized function as somebody that sits in a team and disconnected from a business is not useful, and the article is trying to explain that, but it's trying to get at the bottom of perhaps is a little bit more philosophical.

Loris: [00:37:19] It's because my background is information theory. But if you look at it, Data is just a way to measure the world around us. It's an, you know what results from the act of observing or observing. So we observe and we collect a trace of what we see from that trace. We need to extract information, and that means to learn the new things that we didn't know before. So if you if you look at the definition, the strict mathematical definition of information and information is entropy is is a measurement of how much we can predict the state of the system. So if I was to open a book and read [00:38:00] always the same word over and over and over until the end of the book, that would be a very boring read. And we contain very little information because as soon as I read the first word, I know exactly what the book is going to be about. It's just the same word over and over, whereas a book that and this is obviously an example.

Loris: [00:38:17] It doesn't apply 100 percent, but it's a concept to explain what information is right if the book uses a whole bunch of different words and different concepts. Sentences are not all one liners, but some have long periods, some short periods. My ability to predict what I'm going to read next goes down. And that means the book is is giving me a lot of information is surprising me in a way. And if you look at it, as I'm sure you know, many of the algorithms that we use in classification are based on information theory of light, of looking at the dataset and understand what the entropy literally the entropy looks like regression like classification trees or random forest. I suppose a good example of that. And so if you look at it, the difference between that information this way, the next step is OK, once you've been surprised, once you've read something that you weren't expecting before, so you acquire the information, then what do you do with it? So there's still that extra step and that requires to understand the context around the decision. So I used to I changed my mind mind a little bit because I used to coming from a background on information, theory and telecommunications.

Loris: [00:39:34] It was just about machines, right? You got a satellite up there, forty thousand kilometers, you got an antenna on the ground in 95 96 different beams, radiating power at each other. And you're tasked with that with the, you know, how to how do I design the system so that I can transfer as much information from A to B as possible? But in real life, when we work with Data, there are no machines that are humans [00:40:00] and you can't segment and isolate a piece of work from the bigger context in which the business operates. And that's the challenge that I'm really excited about is connecting the dots is making sure that when we act on a piece of data to extract information, we know how that's going to be used. So I'm really talking about design, and I'm a big fan of the work that Brian O'Neill is doing. I'm a big fan of his podcast as well. That really changed from enriched my understanding of what design is. A lot of people think it's not about esthetics and making things that are pleasant to see, but it's not. Design is about asking the question How are you going to use it? Let's try to make it as useful as possible.

Harpreet: [00:40:44] What's the name of that podcast by Brian O'Neill?

Loris: [00:40:46] I experiencing Data experience Data.

Harpreet: [00:40:52] Definitely have to check that one out. Hopefully, I can talk to the guy. Get him on the show in these days, but I like that. I really that that to me, makes really intuitive sense, right? There's a real world, right? The real world exists as a byproduct of observing the real world. We are able to collect data, but to go from data to information. It's, you know, Data by itself is just like raw rose and whatever, right? Like, for example, you gave you an example about a book like Data in itself could be me just going through and writing the going through every book in my in my library here, just cataloging the books, color, the books genre and the length of pages in each book. And I just have one rule for every one of that. That's just Data in itself. But that in itself isn't really useful, it's when I start creating information and start doing aggregations with that and start saying, OK, well, you've got a bunch of books. The average page count is this your most read genre? Is this so on, so forth? So is that kind of kind of what you meant?

Loris: [00:41:57] Yes, and the first part that [00:42:00] you mentioned is the one that I think is more is it's more challenging from a an organization perspective because as a data scientist, if you have the right dataset and a very specific question and you know what you're doing, you can get an answer, you know, very quickly. So that in itself is a I see it as a single node type of challenge, you know, you have one person that knows what they're doing, they have the right ingredients and they can they can come up with a useful output, but it's never like that. In reality, you know, getting that list of books in a real organization is very, very hard. And it's not a scientist. Or, let's say, folks with an interest in modeling and statistics. We tend to assume that companies have their shit together and they have that inventory. There is a librarian that we can go and knock and ask for what what is what can we trust it? The problem is that's that's not true. We're so,

Harpreet: [00:43:01] We're so naive.

Loris: [00:43:02] We're so naive. We're so behind. You know, the whole information as an asset is a new concept. I'm a big fan of Doug Lenny, and I'm reading his book now in front of me, and I find it extremely insightful because he literally takes all of the stuff that I knew about information from an engineering and mathematical and statistical standpoint. And he sprinkles on top the economic spit, right? So how do you monetize that? How do you think about that as an asset? What does it even mean? Information is an asset. What does it mean? And he and he has this systematic approach and says the information is not deployable. You can have as many information people as possible, simultaneously consuming the same amount of information. And once you have it, it scales. Inherently, it's all like oil and so like electricity. It's in that sense, it's renewable

Harpreet: [00:43:55] And zero marginal cost of consumption, of replication of [00:44:00] use and we're happy with.

Loris: [00:44:02] Yeah, exactly. Obviously, there are costs in costs associated with managing it, like getting the fundamental truth of the universe is that, you know, if you don't do anything. Entropy is going to still increase regardless. So that's why we need to pay the electricity bill for our keep our food cold. That's what the fridge does. Is trying to reduce heat, right? Lower the temperature, reduce entropy. That reduction of entropy is going to cost the money and there is no way around it. That's a fundamental truth of physics. So that's that's true in Data, too, if we want reliable data, if you want to be able as Data. Scientists to rock up and say, I'm going to. I understand the problem, I can take that. And within a day or less, I have an MVP ready. If you want to get to that level of efficiency or fluidity, we need to spend money. We need to spend resources to put order in the chaos, and we need to first start understanding that there is a status of chaos that we don't really know what exists and who owns it. And how is that connected to the rest of the business? So that's where my interest in that management came in after I looked at the modeling, then the architecture. And then I said, OK, we need to go one step lower and start really focusing on the fundamental stuff, which is why what I'm trying to do with that I foundations.

Harpreet: [00:45:27] So let's talk about that, about Data management, so we'll just take my case as an example, since I'm going through this at work, right? So I've got an initiative at work where they're telling me develop a Data management strategy. I don't even know where to start. Right. So I've been trying to educate myself. I've been reading books like Data Strategy, Modern Data Strategy. I've been reading the Data management toolkit, and then I've also got Scott Taylor's telling you Data story.

Loris: [00:45:55] Oh, nice. Yeah.

Harpreet: [00:45:56] You know, these are all been really helpful books for me to to read. [00:46:00] But you know, when it comes to actioning it, it's sometimes hard for me to to start like I can look at all these stakeholder maps and identifying business drivers and things like that, but it's pretty challenging. And you know, if where does one start, right? Where do we start when we're trying to get from the ground up grassroots movement, trying to implement a Data strategy in your organization?

Loris: [00:46:28] I don't have a ton of experience, so my answer is going to be based on a mix of intuition and what I've seen in playing with with change management. And I think the grassroots idea is very important. You need to have people excited that understand the urgency of the problem that really feel the pain of the chaos around them, and they want to do something about it. But as you said that our management is a program, not a project, and it it's all purpose should be to help the business achieve the vision. So I would say that I'm a big fan of Scott Taylor's book. It makes a ton of sense. You got to start from the vision, understand what the senior executives really want to achieve and take that and not start asking questions on the ground to people that know what the status of things is. Do you have the data to to pick it up? Can we can we achieve it? Can we actually do it or do we have to make up stuff? And in that sense, is very much a human challenge. It's about motivating people and talking using that language to explain why data management matters. And perhaps an example I was just listening to. That's the latest podcast on the Data futurologist. And he mentions this example of an airline that, you know, because of COVID. Planes won't [00:48:00] fly anymore, and all they had, they couldn't sell the planes because they didn't own it. They were losing it. And so they went to the bank and said, Look, all we have is this customer loyalty Data, what can we do with it? So they got a specialist to look at it and understand the value of the asset. And it turned out that the dataset set itself was worth three times more the company.

Loris: [00:48:25] So that blew my mind, right? That's what we don't. We don't realize, and I get it because we're not used to think about information as something that we can monetize. But once you take that perspective. It becomes more way more compelling, you can go and talk to a CEO and say, Look, let's let's talk about bottom line, let's talk about what you want to achieve and and let's understand that there is already a ton of value in the data that you have. You cannot do anything with it because nobody, nobody is looking after it. So look after your Data because Data encodes a relationship, no matter how you see it, no matter how you say it. Even if you if you look at two machines and just the data that flows between two robots, those two machines are part of the organization. So they're not they're never in isolation. So there really isn't a case where you can look at a piece of data or an amount of information, and you say this has nothing to do with people as if we are in business. We are. We're trying to do something for people, whether it's, you know, it's our partners, our suppliers or customers. It's all about people in the end. And so caring about Data. And here I'm echoing Scott Taylor. I can hear you in my head caring about that AIs caring about relationships. So that's I think that that's a useful angle. To answer your question, how do you actually do in practice, I think. I don't think anyone has a complete answer [00:50:00] to that. I think starting from the vision is useful and. Once you've got that strong buying from the top, try to champion the communication between everyone involved, and it's really a collective thing.

Harpreet: [00:50:15] Yeah, definitely. Something that popped in my head as you were talking about that story about the airline who found out that their Data is actually worth three times more than the company itself. I think that's because the data itself creates opportunities where maybe they did not exist before. So it has potential for you to uncover new opportunities that could be worth monetizing, whereas having a hangar full of airplanes. That just serves one purpose you can fly from point A to point B, and that's really the only way you can use that particular asset. So whereas Data can be used and leveraged in multiple different ways, so it creates opportunities.

Loris: [00:50:59] Yeah. So the ground zero here is is the literacy

Loris: [00:51:03] Part like we need to make a better job as a community, I think, to educate everyone around the value of that and we have to speak the language of business. If we keep saying we keep talking about entropy in information, that's not

Loris: [00:51:20] That doesn't resonate with the right people.

Loris: [00:51:22] You know me and you. We can have wonderful conversation about the physics of it all and how that connects with with the deepest mysteries of the universe. But ultimately, the word is goes forward based on decisions that are made by four or five senior executives in a room. And if they don't see it, they don't get it. You're never going to get that strong support that you need to implement the program. You know, it's never going to see you as a strategic imperative. It's always going to be perceived as a tax on someone's time. Oh, we need to. We need to look after that. We need to clean that, that those [00:52:00] fields, I mean, you remove null values and people like I really like. My performance is measured on how many recommendation algorithms I can pull together. If you want me to clean up that database, I'm going to lose so much time. And so you know what I mean? Like, it's this very limited, shortsighted mindset that we that I think is hindering the new big challenge of how do we manage the.

Harpreet: [00:52:28] Yeah, I mean, because Data is there to enable you to do other greater things, right? That's been a big thing for me is educating myself and the other aspect the more the business type of aspect of stuff because I mean. To me, coding is fun, doing machine learning is fun, but that's not really what the business actually cares about, right?

Loris: [00:52:51] Yeah, no, they don't.

Harpreet: [00:52:52] So how do we create a culture then for for analytics to thrive in an organization? You know, let's say we're the first data scientist in an organization. We're coming in. We're just like, Oh man, I love code. I love Data. I love creating machine learning models. And then you get in and you're trying to do stuff and then you realize that your hands are tied because there's no. That there's no infrastructure in place, there's no desire or nobody cares about your fancy algorithms and anything like that. How can we start making a culture happen for for success?

Loris: [00:53:26] I think it's a fantastic question and something that keeps

Loris: [00:53:28] Me awake at night. To be honest. I'm really battling with this, and I think the more I think about it, the more I realize that the moment when you are inside the organization as the first data scientist is ready to let. You get you got to act way before that, it's like this huge prison cell cycle. I haven't cracked it. I don't have I don't have an answer, to be honest, but I think we need to speak more business language as that are that are easiest and find [00:54:00] those opportunities to talk about business first and then how that can help the vision that the business has. But we shouldn't shy away from, because one challenge is when you talk to execs, they have very short attention span. They are busy with a billion other things. So the concept of the elevator pitch kind of all around the idea that if you're lucky, you get the chance to stand in the same elevator with with the decision maker if you're lucky and the elevator ride is never more than 60 seconds, and so you get a punch all of that into 60 seconds, I think we should stop thinking about that elevator pitch and more. I recognize that Data is multifaceted, is complex in its nature. And so I'm getting, you know, I have mixed feelings when I hear people making that simple. Yes, you can structure it.

Loris: [00:54:59] I'm all for creating structure, and I'm all for reducing interference and reducing noise. But in a lossless way, we need to be need to be able to communicate the reality of the situation and avoid creating those false expectations because it's all a matter of, you know, I've seen what it is like when people expect you to be the genius that solves all the problems and then you're not. It doesn't feel good for you. It's a waste of time for the business, even if it works. And in my case, we made it work. So that was positive. Happy. Happy ending, after all. It's not an isolated effort, it's not part of a machine, it's not part of this idea of closing the circle of creating that flywheel of your one term and then you go two turns and then four and eight and something that you can do that in that sense, is the perfect medium to implement that [00:56:00] concept of a flywheel because it's not depleting. You do things right today and you close that loop once that stuff, that's going to be useful for some money somewhere else. So one thing I'm really against is creating pipelines is hiding pipelines in in a language that only few specialized people understand and python, as much as we love it and I and I love it. It's still folds in the realm of it, right? You get a software developer, you someone who knows how to program, and that excludes immediately a lot of business people, whereas with things like SQL and I don't want to get into an argument know ETL versus E-ELT, but something like SQL, that's more declarative, not imperative.

Loris: [00:56:47] It's way easier to create to to educate people, to understand it. And that's important because I, as I've done on @ArtistsOfData engineer, whatever you want to call me, I'm going to make assumptions no matter what. We'll have to make assumptions because I don't have a crystal ball and I don't know the full picture as much as I strive to get it and those assumptions, the pillars of everything that we will build in the future. So if those are hidden away and I don't want that, I can look at it. How how do you achieve trust? How do you get somebody to trust what they're saying? They have a dashboard like, I just see a line that goes up, but what does it come from? So we should the tooling that is getting a little better. But I think in terms of the literacy, it really pays off to have people in the organization to have the basic understanding of what or what is involved. And university is not the solution because uni is always 10 15 years behind.

Harpreet: [00:57:48] So let's roll it back a little bit. You're talking about, you know, your first data scientists in the organization. It's already too late. And then we're talking about some of the challenges when it comes to, you know, transforming data and all that stuff. So [00:58:00] you wrote about this latter of Data needs so. So talk about this latter of Data needs that goes from data integration, data access and data transformation kind of walk us through that process and then talk to us about why transformation that part is so hard.

Loris: [00:58:16] Yeah.

Loris: [00:58:16] Well, the concept is very simple, really. Like the first step is if you want to work with Data, it means you want to combine it together. You want to apply some logic and extract information you want to get and actually insight whatever it is you want to do with it, you first need to get the data right. And so that means creating this smoothest path from I need to do something new. I know the business problem. Now let's see what that is available and the moment in which you know what data is available, and you can pull it together and you can start massaging. You blocked it, filtering start to understand it a little bit. And so the integration part is the first step is turning instead of going and knocking 50 different doors to get access to that ElasticSearch from the lead developer and then ask access to that. My SQL database that some, whether it's the legacy system, which still has valuable information, but nobody quite remembers who you know has credentials for it or what is the policy to give you access in the first place. So you have to bag the CEO. Do you think I should get access? Of course you should get access. What are you talking about wasting two weeks, you know, downtime just to have the data? I need the answer.

Loris: [00:59:28] So avoiding all that, all of that and bringing the data in a place where you can go, knock one door, you have access and and consider doing stuff with it. So integration first, the modeling is the second. So once you have it in one place, you have to be able to make sense of it to understand how this together. And so for that, you need structure. I'm a big believer that value is comes from structure. If we just dump stuff in in a in a lake, we [01:00:00] end up, as we all know, with the Data swamp like in an unstructured and maybe it is discoverable, but still so pretty messy because you don't know. What does Rice do you don't know, where does she look at that record or that record? And how do they fit together? So three, and that's the hard part. That's the part where you act as a fringe, right? You are lowering the temperature. You are, you're increasing order, you're removing cows. And that part is very, very expensive in terms of time and brain power. It involves a lot of context extraction. There's a lot of design that goes into that as well.

Loris: [01:00:36] And there's a lot of architecture like asking those low fundamental questions of how do we lay out these assets so that we can keep growing the system we can instead of letting it evolve into spaghetti code where you can't make a change anymore? So a lot of these concepts are very similar to the software development. Yeah, the software development, I think a lot of folks that the love writing software are familiar with the with the concept of an evolutionary architecture. So how do you build that isolation between systems so that you can keep expanding them without losing track of without getting them crazy? And that obviously that the third element is the consumption, so whether that's visualization or or some modeling that really comes downstream, but you have to establish that truth first, then you have to do it reliably. And there's no way to decouple the reliability part from from the business correctness because an analyst would create a model that is accurate, that reflects the business intentions, that reflects the reality of the problem and then answers the problem. But if the whole thing lives in an Excel sheet, it's not going to be useful to the broader organization.

Harpreet: [01:01:56] Thank you so much for that. I really appreciate that. So, Laura Rs2 one final [01:02:00] question before we jump into the random round here. Oh yeah, it's one hundred years in the future. What do you want to be remembered for?

Loris: [01:02:09] Yeah, it's it's a good one. I a loving father. Yeah, yeah. Helper, I suppose. Somebody that. Helped people as much as possible.

Harpreet: [01:02:19] Absolutely love it, man. And definitely, I think you're doing that with your podcast, and I'm excited to to tune in and see your continued success with that. So make sure you guys, I'll include links to all that stuff and we'll get to all your social stuff in a bit here. But I'll include links and everything for you there. So let's jump into the random round, right? So as a physicist, what would you say is the most fundamental truth of physics that all human beings should understand and know? You kind of touched on it a little bit earlier.

Loris: [01:02:47] Yeah, I think it's the uncertainty principle, a very humbling

Loris: [01:02:53] Realization that we can never have a complete picture about a problem or about a situation that we're trying to observe and understand more like accurate terms. The principle really states that you cannot. There are two quantities that are said to not compute, which is a very strange term in physics. But what it really means is not about traveling from A to B. And what it means is, is that you like position and energy or momentum and energy and time position momentum. I'm screwing the equation up because there are two versions, but there are two fundamental quantities that you cannot know with infinite accuracy. So if you know a lot about the position, you're going to have big insurgency on the momentum of the particle and vice versa. And so the equivalent to a position momentum is energy and time. So if you know exactly when you observed the particle, you will have a huge uncertainty on its energy and vice versa, you know exactly the color of a photon. You cannot tell when you saw that photon. So [01:04:00] it's kind of weird.

Harpreet: [01:04:02] That is interesting. I wasn't asking next. What do you think is the most mysterious aspect of the universe, which you say that this uncertainty principle is that? Or is there a different thing that is more mysterious than that to you?

Loris: [01:04:12] I think none, locality.

Loris: [01:04:14] That's not the way we're talking about flipping the state on one side and and seeing that instantaneously on the other. It's kind of a mystery. A lot of people have argued in the past that. It was a violation of the speed of light limit, but actually it turns out that there is no information transpiring, you know, no information is exchanged between those two particles because when you flip one, you know that deterministic, the other one slips through. So you expected 100 percent and the fact that you expect it, you know it already. It means that there's no information. And so the speed of light is still safe. Is the is the fastest the upper limit?

Harpreet: [01:04:57] It's mind boggling to me, man. So when do you think the first video to hit one trillion views on YouTube will happen and what will that video be about?

Loris: [01:05:12] I don't know when it will happen, but I'm sure it's going to be about a cat.

Harpreet: [01:05:14] That's like the most reassuring answer so far to that question. So everybody says some kind of a cat VIDEO. I mean, just buy a bunch of cats now and start filming them.

Loris: [01:05:24] Yeah, exactly. That'd be nice on that part of the world. The white side you have on your left and your left. Yeah, big cat, Mary.

Harpreet: [01:05:33] So who do people tell you that you look like?

Loris: [01:05:36] I don't get that actually a lot.

Loris: [01:05:39] Yeah, no. I don't know. What do you think I look like?

Harpreet: [01:05:43] I feel like you look like a young version of of Walter White from breaking up.

Loris: [01:05:50] Yeah, I can see it. You know, I can see that, you know, I'm laughing because I when I chose these glasses, I actually went by a pair and I looked a lot [01:06:00] like the one that Walter White was wearing in the movie. And my mom is the first thing said. She said of what's up. She's like, No, you look like Breaking Bad. I'm like, Okay,

Harpreet: [01:06:10] What are you currently reading?

Loris: [01:06:12] Yeah, so I'm

Loris: [01:06:13] Finishing in front of me and the next one, I reading insights, which is a book on self-awareness and introspection, which is really, really interesting. And my next one's actually going to be the Kdo Handbook. And the second book that you showed me. In the list, the toolkit from

Harpreet: [01:06:37] A Ernestine beak.

Loris: [01:06:39] Yes, yes, that's going to be my next and I'm also excited about the Data means business and the bee Data literate to other books that are ramping up quite heavily on LinkedIn. Nice. The new reads. Big Fan of Leading Change by John Kotter, which I think is as a classic and. It's it's trying to put a lot of structure into how do you make a change project so successful?

Harpreet: [01:07:06] Definitely. I'll have to check some of those out for sure. I'm actually going to be interviewing Ernestine in a couple of weeks, maybe a month from now. I mean, time goes by so fast for me, so I'm looking forward to that. What song do you currently have on repeat?

Loris: [01:07:21] I am listening

Loris: [01:07:22] To an old album from Alan Tasia, the super band that I think it's from Finland, and they do this progressive symphonic metal type of music. And I used to listen to them when I was a kid, and for some reason I go back into it and I have them on repeat. So there's no particular song but anyone from the album, there's something about it that maybe it's because I'm getting a bit melancholic due to COVID restrictions and stuff.

Harpreet: [01:07:56] Yeah, I'll definitely check that out. Let's go to a random question generator [01:08:00] here. First question on this is what's the last book you gave up on and stopped reading?

Loris: [01:08:08] I don't remember,

Loris: [01:08:09] And the reason is that I stopped reading

Loris: [01:08:12] Quite a few books

Loris: [01:08:13] Because I approached the reading a lot as an experiment. I have my my book bold with many titles, and I get four or five. Periodically, I start reading and if I don't get inspired within the first two or three chapters, I just just let it go. So I don't get that anxiety of finishing the book.

Harpreet: [01:08:33] Yeah, same here. I don't really care if I finish a book or not. I'm just reading it for the idea is just, you know, I'll flip around inside out, upside down, left, right, middle and beginning. Whatever you read, different parts of it. What fictional place would you most like to go to?

Loris: [01:08:48] I'm terrible with this question. Fan fiction plays.

Loris: [01:08:53] Give me an example.

Harpreet: [01:08:54] So fictional place could be like Narnia. It could be like the Shire and the Lord of the Rings. It could be that weird interdimensional place from Interstellar.

Loris: [01:09:05] You want to get to where Avatar was shot, like the planet was avatar in the name of the planet or just

Harpreet: [01:09:12] Forgot the name of the planet. But that wasn't in the movie.

Loris: [01:09:14] Yeah, yeah.

Loris: [01:09:15] That one would be a bit cool to see a bunch of blue blue people going around and flying and huge.

Harpreet: [01:09:20] That's what languages do you speak?

Loris: [01:09:23] Italian and English.

Harpreet: [01:09:25] If you were a vegetable, what vegetable would you be?

Loris: [01:09:30] Oh man, and eggplants, for sure.

Harpreet: [01:09:32] Nice. I love eggplant. Spicy, delicious. How can people connect with you? Where can they find you online?

Loris: [01:09:39] I think LinkedIn is a bad place.

Loris: [01:09:41] I'm on Twitter, too, but it's mostly been awesome.

Harpreet: [01:09:44] I'll be sure to link to your profile and link to your website and podcast and all that stuff. Again, thank you so much for taking time in your schedule to come on to the show today. I appreciate having you here.

Loris: [01:09:54] And it's my pleasure. Thank you for having me.