Rahul Singhal - Clean AI Training Data, and Leading With Selflessness

(Joel Beasley at 00:00:02) Hello, my friends. Today we are talking to Rahul, the Chief Product and Marketing Officer at Innodata, and we discuss how most AI projects fail because of a lack of clean data, how Innodata provides pristine training data for AI algorithms, and why it's important for managers to lead by example and give back to the community as much as possible. All of this right here, right now on the Modern CTO Podcast. Here we go. This is the Modern CTO Podcast.

(Joel Beasley at 00:00:42) Hey, Rahul.

(Rahul at 00:00:44) Hey, Joel. How are you?

(Joel Beasley at 00:00:45) Fantastic. How are you doing, my friend?

(Rahul at 00:00:48) I'm doing well. Thank you.

(Joel Beasley at 00:00:50) Where are you located at? Where are you calling in from?

(Rahul at 00:00:52) I'm in Jersey. I live in Jersey. Where are you guys?

(Joel Beasley at 00:00:57) Down in the Florida area.

(Rahul at 00:01:00) Which part?

(Joel Beasley at 00:01:01) Bradenton. It's about an hour south of Tampa.

(Rahul at 00:01:05) Got it. Got it. Yeah, I was in Miami two weeks ago.

(Joel Beasley at 00:01:08) Oh, nice. Nice. What type of technology were you into when you were younger?

(Rahul at 00:01:13) So when I was younger, I obviously did a lot of COBOL programming and Tandem programming and stuff like that. And then I grew up, I wanted to get into the business side. So I did my MBA from McGill. So I'm an engineer and then did my MBA from McGill in Stockholm and then got into consulting. So I spent a lot of my career doing a lot of consulting work, like business strategy, operation strategy work, majority of it at IBM.

(Rahul at 00:01:42) Then I ended up getting into the Watson Group, which was really interesting. Spent three years at IBM Watson, which was an incredible experience when AI was just being announced, which was kind of cool, right? I mean, Ginni Rometty made an announcement. She's going to spend a billion dollars on IBM Watson.

(Rahul at 00:02:08) Yeah. So I started leading the product stack there, took pretty much what were research assets and commercialized them, then spent two years doing a startup on AI, and then here at Innodata.

(Joel Beasley at 00:02:22) When you were working on Watson, you were there for a while. Was that when they were doing the thing with Jeopardy?

(Rahul at 00:02:28) It was just after Jeopardy. So I was there on the strategy side figuring out how can we commercialize what it was, which was a fantastic experience. And then I came back and did something with Watson from a product standpoint. Yeah.

(Joel Beasley at 00:02:48) That's pretty cool. Was that your very first project in AI?

(Rahul at 00:02:53) Yeah, it was. My first experience in AI, I started with actually a really interesting technology called Personality Insights, where you can use written text and it identifies your personality traits. So we were able to do some really cool, interesting stuff with it. And then we ended up getting into image recognition and applying image recognition and then applying speech models and then applying NLP models.

(Rahul at 00:03:23) So it's great because, again, IBM, to its credit, has always been ahead of the curve. I mean, they called out AI in 2013, 2014 when people weren't talking machine learning. They weren't talking AI.

(Joel Beasley at 00:03:36) Why, like, what puts them ahead of the curve?

(Rahul at 00:03:39) I just think it's research, right? They do really good research. They understand their customers a lot better, and they're able to make some big bets. For example, they called out quantum computing 15 years ago.

(Rahul at 00:03:51) And three years ago, I think they really started talking about it, and it's going to come to fruition in the next two, three years. You will see the reality of quantum computing coming into place.

(Joel Beasley at 00:04:02) Yeah. I got to talk to Robert Sutor. He's one of the leaders of the quantum computing over at IBM. It was one of my first people I got to talk to about quantum computing. He's so smart.

(Rahul at 00:04:11) Yeah. Okay.

(Joel Beasley at 00:04:15) Yep. I like to follow really, really, really smart people. And now I see him like he's been publishing books about qubits and I think the book was called Dancing with Qubits. So it had a nice little humor and interesting, he was just a cool person. Yeah.

(Rahul at 00:04:28) Yeah. Yeah. Oh, I mean, these guys are super, super smart, right? I mean, half the time I can't understand what they talk about, but it is, it's fantastic, right? So I think, to IBM's credit, they call things out. I wish they would execute better, but I think they do call out the trends sooner than others.

(Joel Beasley at 00:04:55) Yes. And they're a very large organization. So, I mean, the more mass you get, the harder it becomes to execute.

(Rahul at 00:05:02) It really is. It takes a lot to execute. I mean, it becomes harder, especially when you're a publicly traded company, right? When you're trying to balance innovation with quarterly growth, which is hard.

(Joel Beasley at 00:05:16) Yeah. I was talking the other day with an Army general, and

(Rahul at 00:05:20) Mhmm.

(Joel Beasley at 00:05:20) He was also into technology as a CTO. And he mentioned that one of the reasons why the U.S. Army is so successful is because they push down the decision making. And it makes it really hard to defeat the enemy when you empower people at the lowest level to make decisions based on the current context.

(Rahul at 00:05:39) So there's,

(Joel Beasley at 00:05:40) there's a map for, you know, this procedure, how to flank. But given localized context, they will not even follow their own plan given down from up above due to their empowerment to make those decisions. And I was like, that's such an interesting perspective as to why other countries say the U.S. military is, you know, more than adequate.

(Rahul at 00:06:04) Yeah. That's right. Right? I mean, you've got to empower your people. You've got to decentralize it, and then somehow you've got to bring the command and control back, right? Yeah. You can't let it be all, so you've got to bring it back.

(Joel Beasley at 00:06:18) Yeah. I know. It's definitely, it's a fluid motion, right?

(Rahul at 00:06:22) Right.

(Joel Beasley at 00:06:22) So tell me a little bit about what you did after Watson.

(Rahul at 00:06:26) Yeah. So one of my customers at Watson was a company called Equals 3. They rebranded themselves as lucy.ai. And it was started by three co-founders, three founders who'd been in the martech industry for 15 years, very successful. And they were looking to solve a problem of needle in a haystack, right? So they wanted to say there's so much unstructured content out there for marketers, right? When you are a marketer, you have access to millions of pages of PowerPoints and unstructured content. Then you have first-party data. You have third-party data like Gartner and Forrester and other kinds of subscriptions. And you spend more time trying to find data than actually adding value, right? So as a marketer, you might want to understand, if you are creating a digital campaign, hey, how many people live in, how many people who are African American like toothbrushes in Brooklyn with zip code 07078, right, or whatever.

(Rahul at 00:07:34) That data will probably take you half an hour to 45 minutes to go to an MRI database and go find the answer. So they would apply machine learning techniques to find that answer in actual language. So I joined them as the Chief Product and Customer Success Leader. We built out a product which won several awards, including IBM Watson awarded an award for the best ISV, and HubSpot Innovation Award, and very successful in terms of trying to solve the problem of a marketer using machine learning and deep learning techniques to extract metadata.

(Joel Beasley at 00:08:12) That's pretty cool.

(Rahul at 00:08:13) So I—

(Joel Beasley at 00:08:13) I could just tell the system, hey, I want this information. I type in a text box.

(Rahul at 00:08:17) Exactly, right. So you ingest the data in, and you type it in just words, and it'll come back with top 10 answers. And you can then query and go all the way down. So it went really, really well. We raised, at that time, $8.9 million and continues to do really well.

(Joel Beasley at 00:08:34) And so then you move on from that to Innodata, or was there some projects in between?

(Rahul at 00:08:39) No. I moved on to Innodata. And Innodata approached me. It's a really interesting company. The heritage of the business has been we've been around for 30 years, publicly traded, and 3,500 people around the world with really deep subject matter expertise in legal and publishing, health care, and financial services. And what attracted, so at that time, two and a half years ago now, the CEO and the board were looking for a Chief Product Officer to come in and really transform the business.

(Rahul at 00:09:10) And one of the things that really attracted me to Innodata was, having spent by that time seven years in the AI/ML world, my hypothesis and thesis were companies that are going to be successful are ones that are going to be either solving a business problem using deep learning, machine learning techniques that is applicable to a business problem, or companies that have access to proprietary content, or companies that have really deep subject matter expertise and are able to create content, right? Digitize content and create content. And Innodata is a company which is really exciting because we have people who understand content at a very granular level. So we have doctors and physicians on our staff. We have statisticians. We have lawyers where we are their middle arm of creating content for them, which is sold as a digital product. So my thesis coming in was, you know, it is, it's an exciting company with AI. Again, another big thesis is AI is going to become part and parcel of every application.

(Rahul at 00:10:33) Like, companies will not have a choice but to apply AI technique to solving a certain problem because that's going to be expected by your customers. To solve those kinds of problems, you need very clean, pristine data sets. Having spent a lot of time with Jack Abuhoff, the CEO is extremely well positioned to take advantage of the growth trajectory that was, that is coming and that thesis is starting to play out very well.

(Joel Beasley at 00:11:02) So when people ask you what you do at Innodata, what's the short response?

(Rahul at 00:11:07) I empower my teams to help, and my customers, to help create training data so that they can go build sophisticated algorithms to solve business problems using machine learning and AI.

(Joel Beasley at 00:11:24) Brilliant. Yeah. That's amazing. That's so cool. I'm excited.

(Joel Beasley at 00:11:29) I usually don't talk this much, but you got me all excited because I want to know, we're way, I feel as if we are in the times when computers were the size of a room, in relation to AI. It's just in its infancy. And I'm excited for, you know, the iPhone. And so I'm living in this past, and I'm excited about this future. And the way that would relate to AI is, I think it's going to be cool when there's cyborgs walking around, and they're indistinguishable from humans unless you perform some sort of identification function on them or something. That'll be cool.

(Rahul at 00:12:04) It is. It's going to happen, right? I mean, it is going to happen. So everything that we, so go back, go back to Star Trek days. I don't know how old you are, but go back to Star Trek days. How did you ever imagine they used to open up that thing and they would have a video conferencing with people below, right? Earth versus in their, upstairs. That's FaceTime for you.

(Rahul at 00:12:28) Right? Now I can go and talk to my parents in India. I can talk to anybody, talking to you face-to-face as if I'm sitting next to you. Advances in technology are happening at astronomical pace, and I think AI is absolutely going to transform businesses.

(Joel Beasley at 00:12:48) Yes. And it's interesting because we tend to take inspiration from our media and make that into reality. Not everything makes it, right?

(Rahul at 00:12:58) Right.

(Joel Beasley at 00:12:58) But when things are realized multiple times in multiple different ways, you can look at, you know, you got Star Trek, but you also have the Jetsons. They're all talking to each other on these screens, and then that ends up becoming reality. And so I'm always interested in, well, let's go back 20 or 30 years from now into the movies, or let's look at what's happening right now today in the imagination and the movies, and that's going to help us predict what's possible in the future. And I was like, that is a cool aspect of the human species.

(Rahul at 00:13:26) That's true. It's pretty neat.

(Joel Beasley at 00:13:28) So okay. So you guys clean this data, right? What do your customers look like?

(Rahul at 00:13:32) Yeah. Great question. So our customers are anywhere from startups. So we do a lot of work with small companies, seed capital, Series A, Series B, Series C. I'll give you a few examples. One of our customers happens to be a company which is looking to automate claims processing in the auto insurance world, right? So today, when you think of the business problem, you have an accident, right? You take some images. You have a claims adjuster come in. They go take a look at your damage. They send it out to their underwriter who takes a look at it. They make a determination. Hey, should I replace your windshield, or should I replace your fender bender? Or do I correct it and then they give a quote, and then you go make those changes? The problem they're looking to solve is they're looking to say, I can take an image. I can load it onto the app, and the machine will identify the kind of damage it is, identify the car model. They're able to connect to their system, underwriting system, and auto-adjust it and give you you're eligible for $5,000, right? And you can go do something, right? So they save so much money because on average, it costs them $150 for a claims adjuster to come in.

(Rahul at 00:14:57) Now you're automating that whole process. You're turning around the time. So those are startups. They raised $3 million, $4 million. We are working with some very other interesting startups in, say, autonomous cars and image recognition models and some with very large social media companies and machine learning companies where we are helping them with creating models for toxicity analysis or helping them understand what toxicity analysis is. Hey, is a certain content toxic? Is it violent, right? How do you automatically understand that content? We are helping companies, some companies identify different objects, right? So we have worked with some very large companies where they give us 2 million, 4 million different images, and we are annotating the data so that they can train the models to identify an object. And then we are working with banks. We are working with a large number of banks where we are working with things like ISDA contracts, which stands for derivatives contracts, where we are taking any unstructured contract, and we can apply machine learning techniques to extract metadata. So it goes all the way, long answer, from Series A to large tech companies to large banks and health care institutions.

(Joel Beasley at 00:16:22) That is so cool.

(Rahul at 00:16:24) Yeah. It's been fantastic to be able to see the kind of use cases and applied AI that is happening today. We get to see in a variety of industries, and every day I get blown away. As an example, I just got a lead from a car manufacturer where they have large amounts of drone imagery where they want us to look at, they want us to create training data from the drone imagery, which is video data, so that they can create the right path for acceleration from an autonomous car perspective, right? So that's very cool, right, to get involved in those kinds of projects. We are working, I can't disclose the name of the company, but we are about to start a project for autonomous cars where we are going to be helping researchers in building all the training data required to create, for self-driving cars to happen, right? And you can imagine the amount of data required to solve that problem.

(Rahul at 00:17:36) So we are about to, it's called LiDAR data, which is a specific format. But so we are starting to work on those kinds of problems now. Yeah.

(Joel Beasley at 00:17:45) I learned about LiDAR when I was listening to Elon Musk give a talk. Yeah. I won't go there. I won't go any deeper. Yeah.

(Joel Beasley at 00:17:53) That was a lot of fun. But it's so interesting to see the different ways because, you know, who's not geeking out about autonomous vehicles? It's so...

(Raul at 00:18:00) Very cool.

(Joel Beasley at 00:18:01) And there's so many different ways to solve a problem. That's the interesting thing about engineering too. I used to have a specialty consulting where people would come to me, and this was, like, in hindsight. But people would come to me and they would say, alright, here's a patent on a software. We want to perform the same function better without violating the patent. Make it happen.

(Raul at 00:18:21) So that...

(Joel Beasley at 00:18:22) was like, there's a challenge. But because it's actually pretty easy, there's just so many different ways to solve a problem, which made me kind of start to question why investors put so much weight on patents. I was like, I don't know if you guys know this, but they can just find a smart person who can figure out how to achieve the same result in a slightly different way. It's possible.

(Raul at 00:18:46) True. It's a business.

(Joel Beasley at 00:18:46) Yeah. I'm sorry. I'm off topic. Okay. So you do this AI training data, right? You've got these amazing customers and fantastic customer stories. How do you keep up with, like, how do you choose where to spend your time? If you're a large company, you have lots of customers, lots of stories. How does a customer make it to you?

(Raul at 00:19:07) Great question. So we do a lot of advertising. We spend a lot of time in content marketing and, obviously, paid search. So we have a fantastic marketing team where our goal is to really democratize AI, right? We truly believe that companies today are struggling and not being able to solve problems with AI because they lack clean training data. It's as simple as that. It's the old adage of garbage in, garbage out, right? You don't have good data. I'll give you some statistics which are really interesting, right? One is Arvind Krishna, who's now the chairman and CEO of IBM. I think a year and a half ago, two years ago, publicly came out saying 85% of IBM projects are failing because of lack of clean training data, right? Michael Conklin from CIA, who's the head of data there, came out with very similar statistics, right? So it's a well-known fact. Not having access to good, clean, annotated training data, you will fail in your AI initiative. And if you think of the amount of money being spent on AI, it's expected to be $125 billion in the next five years. If 85% or 90% of those projects are failing, imagine the amount of money that is going to waste. That's $100 billion that can feed countries. So long answer short, we want to educate people. Our marketing approach is by writing and talking about the types of problems we are helping our customers with so that they understand and they know that there are companies like us around that could help them as they go and deploy their AI solutions.

(Joel Beasley at 00:21:08) That's the smartest move. I'm telling you, it works. It is the current great move to be the one educating. Be the YouTube video people first see when they're exploring the topic, right? Because they're gonna start learning from you. They're gonna learn your background, and then it just becomes, you build trust. You build trust and respect on the topic. And then when they need to go to the professional services route, when they need to start spending money in that area, you're the person that they know and trust when you've never even had a conversation with them yet.

(Raul at 00:21:43) Yeah. So I think, and we also have a great sales team, direct sales team, which goes and tries to talk to data scientists and the CTOs and educate them on what we can do and how we can support them. And the problem is actually very simple. The problem is enterprises who are spending hundreds of millions of dollars on AI and ML, they can't expect their existing workforce to create this training data because everybody's working 120%, 150%, right? So this work is over and beyond what you have to do, and it's not a fun job. It does require subject matter. It's pretty easy. It does require transforming the data into a certain format so that the machines can understand in a certain format that it can be taken. And what happens is there is data drift. Data drift is, like, data scientists, they build a model. They find, oh, you know what? I forgot a certain field. Maybe this field needed to be annotated. So you have to go retrain and create new training data. Oh, by the way, the data changes, right? Just the parameters of the data change or you have to retrain the model. So it becomes a continuous activity. So when organizations who are embarking on these kinds of initiatives, they really have to, we urge our customers to allocate a substantial amount of the budget thinking about how are you going to create this training data. How are you going to manage the data? How are you going to manage issues around data drift? And allocate budget and be thoughtful about the full AI strategy before getting started.

(Joel Beasley at 00:23:22) Are you guys a publicly traded company?

(Raul at 00:23:24) We are. We are publicly traded.

(Joel Beasley at 00:23:26) Okay.

(Raul at 00:23:28) We just had our first quarter earnings, and it was fantastic, right? We're starting to see effects of the work that we've been doing for the two years starting to showcase in revenue growth.

(Joel Beasley at 00:23:40) That's exciting. The way I'm looking at this is you guys are like oil, and people haven't even gotten all the cars yet. And in ten years, you're gonna be significantly larger than you are. This is what everybody needs. It's like the prerequisite to the AI is the data. It's a smart business, so that must mean that there's other companies doing this, right? Like, you aren't the only one.

(Raul at 00:24:01) We aren't the only one. There are small and large companies looking to solve similar problems.

(Joel Beasley at 00:24:09) And how do you differentiate yourself?

(Raul at 00:24:12) Yeah. Again, I think there are two big reasons, right, how we differentiate. The first one is quality. So our company's heritage has been digitizing content for information publishers, right? So if you think about a company like Bloomberg or a company like Elsevier or any of these large companies who produce digital content, they require 99.99% accuracy because that data is being used. Think about Bloomberg. That data is being used to trade, and you can't have any mistakes in the data. So our quality processes are so well tuned to that kind of quality level, which means we have teams across the world which are InnoData employees, and we train them, we staff them. We have very rigorous quality parameters that we look for when we give our final deliverables. We are winning work from a lot of our competitors because companies are coming back and saying, hey, the quality level wasn't great, and we really need someone who can produce 99% accuracy. I'll give you a simple example. Just in the morning, we won a million dollar project from one of the largest tech companies in the world, and we are going to be scaling up to 100 people to begin with. And the chances are it'll scale to 500 people. But we're taking away work from an existing vendor who could not produce that quality level. They were producing 95% quality level. They expect 99% quality level. That quality is absolutely the number one differentiator. And if you have great quality, the other thing which we have is we build our own AI ML platforms that allow us to, at a large scale, create this training data. We have our own platforms. We think we have some unique differentiation around our ability to, on the tech side, be able to take, convert any kind of textual data into machine-readable format and then pass it through large teams that can do the annotation and then produce the right quality metrics. Like, quality metrics that data scientists look for are things like Krippendorff's alpha, or they look for metrics like kappa metrics. These are important metrics to understand whether high quality data has been created and prepared. Our platforms provide these kinds of metrics on the fly, giving access to the data scientists comfort that the data that they're gonna be training the model for are really good. And finally, we build machine learning models that can auto-annotate, right? So think about the amounts of data you need to be able to have machines auto-annotate the data. So we also have a platform that is a no-code platform that no data scientist is required and the model starts to get created as we create the training data and you accelerate the training data creation.

(Joel Beasley at 00:27:13) So cool. So you are the difference between 95% quality and 99% quality.

(Raul at 00:27:20) Absolutely.

(Joel Beasley at 00:27:21) I love that. I've always wondered when I first got started making businesses, I thought to myself, why do some people wake up and they choose to be, like, the Bentley of cars and other people wake up and choose to be, like, the Toyota of cars. And I just didn't really understand it. Now I find it's a lot about, like, principles and culture, things of that nature. And so whenever I'm shopping for a product, I start thinking, I take those things into account. Like, what's the culture like at this company? Like, who are these people? What do they believe in? And that helps me understand, like, what's going on behind the scenes there. Hey, I want to take a break from the AI stuff real quick. I just want to know, like, what's your, you have kids, families? What is your...?

(Raul at 00:28:07) Two kids, 15 and 10. So a high schooler and a fourth grader.

(Joel Beasley at 00:28:16) Boys, girls?

(Raul at 00:28:17) Boy and a girl.

(Joel Beasley at 00:28:18) That's exactly what I have. Yeah. That's exact. I have a boy and a girl. The girl's about four. The boy is just over two. Very cool. And so I just approximate, right? She's just about four. He's about two. We're out of the counting months part, right, when you're doing, like, the eighteen, twenty-four months. But, man, I'll tell you what. It is so rewarding, and honestly it makes me think that we're just really advanced computer systems, Raul, because you watch them learn and you watch them go through these training sets and you watch them pick up data and you see how they're learning and they're little AIs, and it's brilliant.

(Raul at 00:28:56) That's right. That's right. I mean, it is, we are about to get a dog. And that's gonna be an amazing, interesting experience on how do you train a dog. Like, are there experiences and techniques of, you know, because it's all about rinse and repeat, right? You teach the dog you have to do a certain thing, and you hope that that's what the dog is going to continue doing.

(Joel Beasley at 00:29:27) And the hard part's the discipline. Like, it's easy to know what to do. I think it's fairly easy to figure out what to do through books or conversations with mentors. I'd say the hardest thing is doing it consistently over a long period of time. That's difficult. But you've been able to do that. That's why you've succeeded in your career.

(Raul at 00:29:47) I hope so. Yeah. I mean, I think it's been a great journey. Can't complain.

(Joel Beasley at 00:29:52) Where do you think your discipline and drive comes from?

(Raul at 00:29:55) I think a lot of it comes from my parents, right? I think my dad's probably one of the hardest working people I've ever known. He's a professor in India. And he's written over 200 books, and he continues to still get up at 4AM in the morning and continues to write and teach. So I think it's something that's really helped me, helped my father. And then just seeing him do it, I think, has helped give me the discipline to, you know, work hard. It's about hard work and work ethic that drives success. I firmly believe it's not about intelligence. It's really about hard work and work ethic.

(Joel Beasley at 00:30:39) Oh, I can tell you it's not about intelligence because I'm not that smart of a person, and it just takes a long time of hard work.

(Raul at 00:30:47) It really is, right? It's perseverance. It's absolutely, and I tell this to my kids all the time. The winners in the, firstly, it's a marathon. It's not a sprint. So life's a marathon. It's not a sprint. Remember that. And number two, remember that it's all about being able to take any job and doing it to the best of their abilities, right? And then just persevering. And if you can just have those three things, you will do fine.

(Joel Beasley at 00:31:14) Absolutely. Yeah. It's kind of hard too because, like, physiologically, we have emotions, and those are typically short spurts, right? You'll go through moods throughout the day. And so we're sort of trained on, like, these shorter cycles. And to have success, you have to have these longer cycles. It's, I don't know, pretty interesting. Okay. So you do all this work, right? You get to where you are. Did you say that your dad has written 200 books?

(Raul at 00:31:39) Yeah. Yeah. He's written over 200 engineering books, which is mind blowing to me how he's done that.

(Joel Beasley at 00:31:45) Have you written any books?

(Raul at 00:31:46) I haven't written yet. Yeah. It's been a goal. I just haven't had time or the drive to do it. So one day.

(Joel Beasley at 00:31:58) Do you have the same name as your dad?

(Raul at 00:32:01) Last name. Yeah.

(Joel Beasley at 00:32:02) Oh, okay. Because if you had like, my brother and my dad, they have the same first and last name, right? But if you did have the same name, you could, like, take a little bit of credit for some of those books. Not all 200. You get credit for maybe, like, five of them, right?

(Raul at 00:32:14) I would take that any day.

(Joel Beasley at 00:32:16) If you did write a book, what would it be on? Do you think it'd be on leadership or AI?

(Raul at 00:32:21) That's a good question. I haven't thought about it. I probably, so I teach at NYU. I've been an adjunct professor at NYU for seven years, and it might be about, well, life experiences, I think. It might be about, you know, it won't be a business book. That's what it is. I think it'll be about how do you take your experiences, and can it help people in some way or the other.

(Joel Beasley at 00:32:49) I like that. Because it's like giving back. It's, you're trying to help the next generation do better.

(Raul at 00:32:55) Right. Right. I think it would be anything that I can contribute back. So it's what I would probably want to write about.

(Joel Beasley at 00:33:03) So how do you go about, you know, spending time and contributing to your direct reports so they pick up on these behaviors and habits that you've learned?

(Raul at 00:33:11) So one, I think the example that you gave is something that I firmly, from my management style, it's about decentralization and empowerment. So I've always, I've always believed in, you set the direction. You give them the goals. You give them the strategy, and then you let them go do it. And then you measure them on very clear KPIs that have been driven. That's one. So decentralization and empowerment and then giving back to the society, right? I mean, I think you lead by talking about, like, for example, one of the things we are, you might be seeing, India is going through a very horrible time with COVID, second wave of COVID, and we just launched at InnoData, we launched a charity drive. We are helping our team members in India and all around the world with oxygen concentrators, where we are providing concentrators because there's lack of oxygen cylinders and concentrators and ventilators.

(Raul at 00:34:15) So we are raising capital and then figuring out a way around the world to get to India. Right? So I think you just find ways to lead by example, and hopefully people will follow.

(Joel Beasley at 00:34:26) Nice. Yeah. I saw that Marc Benioff, like, got an airplane and filled it with supplies, and then a couple other tech people did some things. It's great to see the world coming together to help each other. I love it.

(Raul at 00:34:38) It's been great to see the whole world coming together and helping out in the time of crisis. It really is a humanitarian crisis in India right now.

(Joel Beasley at 00:34:48) So is your dad still over there? How's he doing?

(Raul at 00:34:51) Yeah. He's fortunately, actually, they're doing fine. But I have friends and a lot of family who has gotten COVID. And the horror stories just with the sheer number of people who've gotten COVID and inability to find hospital beds and oxygen at the right time is hurting. So, yeah, we're doing everything we can from here.

(Raul at 00:35:21) That's hard. Right? I mean, it's really hard because there's only a certain amount of infrastructure that's available to support.

(Joel Beasley at 00:35:28) Have you gotten to spend any time over there?

(Raul at 00:35:30) So yeah. Yeah. So I go there every six months to a year. I try to go visit them or they come visit us. And then we have large teams in India, so I try to go spend time with my teams in India every six months.

(Raul at 00:35:42) Unfortunately, for a year and a half, I haven't been since COVID started, but hoping in six months we'll be able to go visit them.

(Joel Beasley at 00:35:51) And have you done special online events or something to make up for that? How have you handled that?

(Raul at 00:35:55) Yeah. So we do special events. We try to host monthly chats or Zoom calls now with our teams around the world and then try to make it a lot more fun. So we've done some events with the team. Gets harder with the time differences, but we try to manage it and try to have a monthly, bi-monthly events, which is a fun event that you can do, like trivia or an escape room or do wine tasting, those kinds of events within the team just to keep the morale up and going.

(Joel Beasley at 00:36:29) And you seem like a pretty charitable person. This thought's been going back in my mind a couple times throughout this conversation.

(Raul at 00:36:36) I think I can do more. I think I can do a lot more than I do, but yeah, I think I do believe we should give back to the society as much as we can. Right? Whatever you can to the best of your ability, give back to the society.

(Joel Beasley at 00:36:52) At what point in your life did you figure out that this is a good thing and giving back feels good and this is something I want to do more of?

(Raul at 00:37:00) I think I've seen that just growing up. Again, from my parents, I've seen them always taking a certain percentage of the paycheck and giving back to charity. I would say I was not as charitable as I should have been in the early part of my career, maybe first ten years, and then I think slowly got it back. And now it's become a really important aspect of our lives.

(Joel Beasley at 00:37:30) Yeah. My wife and I, we give back a percentage of our income to the local community, and we didn't start doing this really until about a year ago. But I'll tell you what, it makes you feel different when you're contributing to your community. You feel a sense of ownership and pride and responsibility for the things that are happening. It's a whole other level.

(Joel Beasley at 00:37:52) It's different than taxes. Taxes are one thing. Right? But giving above taxes and just contributing to the community, it selfishly makes you just feel really great.

(Raul at 00:38:03) You're right. I wish what I would love to be able to do is give more time than just money. So that, I think, is the next aspect of giving back. Because I think it's easier to give money. I think it's harder to find time, and that's the goal is how do we give back time in whatever way. Right?

(Joel Beasley at 00:38:23) Yeah. No. It's good. I figured it out in my late twenties, my first experiences with it, and then just slowly has gotten more serious into my early thirties where I'm at now. So it's something I'll see myself doing for a long time.

(Joel Beasley at 00:38:37) But you're right. You have to find that balance. Right? You can't just quit your job and go full time serve the community. You just have to find that balance.

(Joel Beasley at 00:38:45) We've got about ten minutes left. Two other topics I want to talk about: a little bit about leadership, and then I want to give you some time to wrap up and, like, calls to action for people reaching out to you. So my question for leadership is, if you could design the perfect leadership training program for your direct reports, what would the two or three things be in that program?

(Raul at 00:39:10) The one would be absolutely how do you lead by example. Right? So what kinds of things that you would do to lead by example? So how do you do things? Second would be—we talk a lot about charity. I think every leader has to have a charitable, in whatever format. It really doesn't matter how you do it, but you have to be able to give back to your community, to your people, to the society in some shape or form. And then would be your management style. Right? How are you managing people? And there's a book that I love. It's called "The No Asshole Rule." I make everybody read that book. How are you not an asshole. Right? And I pardon my French on this one.

(Joel Beasley at 00:40:06) No worries.

(Raul at 00:40:07) But it is, it's so important to me. I've come across so many smart leaders who just would be really rude and harsh to their people. I don't think you can drive performance by screaming and being hard and being a jerk to your people. So how do you—so one of the things I look for when I hire people and leaders in my case is, does he have, what's his likability quotient? Right?

(Raul at 00:40:39) So one of the things that I personally measure people on is the likability quotient because that's what really gets you the call. Right? That's what gets you to go. So I think those are the three things we look for, and you should interview my boss. He's probably one of the best leaders you would ever come across because he—

(Joel Beasley at 00:41:01) Really?

(Raul at 00:41:01) All of—oh my god. He is, he's amazing. So—

(Joel Beasley at 00:41:04) What's his name?

(Raul at 00:41:05) Jack Abuhoff.

(Joel Beasley at 00:41:06) I'll have my team reach out and follow up on that too because I love great leaders and learning about what they do and how they work.

(Raul at 00:41:13) Fantastic. I mean, he's one of the best leaders I've ever worked with, worked for.

(Joel Beasley at 00:41:19) So let's wrap up with some calls to action. Why would people reach out to your company? What problem are they experiencing that you can help with?

(Raul at 00:41:28) Yeah. I think any company that is deploying AI solutions and are looking for accelerating their AI because they don't have enough clean training data. We are the company that can actually help across image, video, speech data, and help save you guys money.

(Joel Beasley at 00:41:49) Cool. Nailed it.

(Raul at 00:41:51) So how—

(Joel Beasley at 00:41:51) How do people reach out and find out more about you?

(Raul at 00:41:54) They can go to our website, fill in a form, or they can reach out directly to me or my team.

(Joel Beasley at 00:42:02) And what's the website?

(Raul at 00:42:04) www.innodata.com.

(Joel Beasley at 00:42:09) Perfect. Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague that you think would get value from it. And if you have topics that you would like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.