Evolving Industry:

A no BS podcast about business leaders who are successfully weaving technology into their company DNA to forge a better path forward

The Age of the Generalist: Why Critical Thinking Will Be Our Most Powerful AI Tool

George Jagodzinski (00:02):

What can we learn from teaching the next wave of college students data and AI? What interesting challenges and opportunities are unearthed when 25 years of data leadership meets a young and eager class of AI students?


Today, I'm joined by Kishore Aradhya, who has led data teams and organizations such as Staples, Bose, Stanley Black & Decker, and many, many more. He is currently teaching classes in data engineering, gen-AI, and large language models at Northeastern University. He shared tons of his insight with me. Please, welcome Kishore.


Welcome to Evolving Industry, a no-BS podcast about business leaders who are successfully leading technology into their company's DNA to forge a better path forward. If you're looking to actually move the ball forward rather than spinning around in a tornado of buzzwords, you're in the right place. I'm your host, George Jagodzinski.


Kishore, thanks so much for joining me.

Kishore Aradhya (01:11):

Happy to be here.

George Jagodzinski (01:13):

Something that I was really excited about to explore with you because you're at a really different… you’re at a really interesting point in that you've got 25 years of experience behind you in data, analytics, insights, AI, ML automation in the industry. And now, you're teaching students, and I'm curious to explore a little bit of, what's the cold, hard reality that's hitting these students when they come in versus their maybe idealistic views of what might be going on with AI?

Kishore Aradhya (01:44):

Yeah, that's a really interesting question because I always start off with the students, trying to kind of extrapolate and trying to get down to the real deal in terms of that, meaning a lot of them come in saying, "Oh, we're going to build this model. Oh, we're going to do this. Oh, we're going to use this new tool," or whatever new tool of the day is. And I always go back to them saying, "Look, at the end of the day the most important thing is data, the right kind of data, is the quality of the data that you have. And is the data the right fit for the business question that you want to go after?"


So I always start off with them saying, "First and foremost, before you dig into any kind of either data analytics engineer or data science, you really need to first understand what problem are we trying to solve. Once you get a decent handle about that, only then do you start to look it up and try to figure out the right kind of domain. Do you have the right kind of knowledge, more folks to help you with that?"


So that's kind of where I go to start off with that.

George Jagodzinski (02:46):

I love that. I know we're always preaching that, you know, people kind of... Just like a golfer likes to buy a new golf club, organizations tend to just buy technology for technology's sake. And you have to go back to, “What is it we're actually trying to solve here?” I'm curious of some of the techniques because these students, I'm assuming, mostly don't have a lot of that time in the trenches to have that pattern recognition of what problems. What are some of your techniques to get them more problem-oriented?

Kishore Aradhya (03:13):

Yeah. I always start off by saying, "Look from the first principles." I mean, one of the things that I've had to do that is to show Matt Turk's, this big thing where there's so many different vendors, if you will, like thousands of them. And I say, "One of the challenges, as you enter the workforce, no matter where it is, everybody will have their own pet tool, if you will, that they want to use, or that they would like to use. You should always question everything in terms of what problem is it trying to solve. What is the basic, fundamental need of not the tool, but the problem it is trying to solve."


Like, if you're looking, for example, data transformation techniques. There are so many tools out there that does that, like Matillion. You name it, there's so many of them that are out there. The question then becomes, can we do that minimally with a fewer amount of these integration pipelines, fewer amount of these technologies bringing into force? Can you first do it using simple Python Scripts, for example? Can you use it using with Airflow? Airflow, you need in this orchestration, too.


A fix on one or two of these, too, and see if we can get the same thing done without any of the other abstraction layer around that. So I always go down to the first principle saying, "What is it that you're trying to do? And, try to minimize the number of interactions of different techniques you bring into it." Always start... There is an older self-ranging principles is the best code written is the one that's not, at the end of the day, because it's less maintenance, less things.


It's the same thing that needs to apply in the data engineering world as well.

George Jagodzinski (04:53):

It made me laugh because I'm sure you remember the days when productivity was measured on how many lines of code you were cranking out, which is just the silliest thing in the world.

Kishore Aradhya (05:00):

It really is. And I always try to get them to use our managed services, used server lists. Use things that have already been built, that have already been hardened. Don't try to invent those things. Create value further upstream to the business. Do not try to do that at the downstream level because nobody really cares. Honestly, nobody really cares about if you've written the fanciest code or not. It is to provide value to the business in a reliable way. That's all they care about.

George Jagodzinski (05:28):

Yeah. That's fantastic. And then, when we talk about AI, we can all be grumpy about there is a lot of buzz versus reality and what's real. What I'm interested to hear from you is you're dealing with real students, preparing them for real jobs and what's next. What are you finding, this may be heartening and disheartening as far as what you're seeing, as far as AI and the students?

Kishore Aradhya (05:56):

Yeah. It actually takes on an interesting note. Some students are more tuned to that than others. Those are more tuned to it somehow believe magically that using either Copilot or any of the data… whatever it is helps them to kind of not focus on the fundamentals, not focus on the things that leads them to get there, right? That's one end of the spectrum.


And the other end of the spectrum is people are just like, “Well, this is not going to change our life. It's not going to be that,” which is surprising to me that people would actually think that way. It is going to change, and one of the things that I've tried to get them to see is use these tools. That's what I tell them. Use them in the class. I expect you to use them because there will be others using it. But the most important thing there is, to understand that whatever be these calf holding, whatever be the code that gets spit out, you need to understand. Is it the right fit? Is it the right place for it to be used, using these libraries or using these routines?


Because every organization has certain basic approaches to building capabilities up. They have certain principles, they have certain approaches to using code formatting in a certain way, or what have you. There are certain things. Like with Python, of course, you use the standard… there are certain mechanisms piled in. There's mechanisms that you can do that. So learn to figure out what that is so that when you use these systems at play, these Copilots of the world at play, that you know to discern what should be used or what should be ignored. So that's the way that I say it. Use the tool, but use it in a critical way.


Just like critical reading is important, critical thinking is important. You need to know the fundamentals, other than just to let the tool do the work for you and think it's going to solve the problem.

George Jagodzinski (07:47):

Yeah. That makes a lot of sense. You know, one thing, I feel kind of out of touch with that age range right now because my child, she's six years old, and I'm now mid-40s. And we don't have a lot of fresh-out-of-college employees at our company. I'm curious what other general challenges are you seeing? One I can think of is how many times have we been in an organization where you try to solve a problem, and you've got some people that have been there for a while, and they say, "Oh, we've tried to solve that before. You're not going to be able to solve it."


And then, you combine that with these more junior people and new technology in the mix. I'm curious if you're seeing that, if you have any kind of techniques and approaches on how to overcome that?

Kishore Aradhya (08:27):

Yeah. It's an interesting thing. We're entering into a generational shift, if you will, in terms of… and I think we've all seen this, I've been doing this for over 25 years. So over the years there's always been the folks who have been there, done that. I've seen this happen. And there are folks who are entering the workforce who kind of feel that they can bring in these new approaches to doing it. I think there's value from both sides of the equation.


So what I have done in a lot of organizations when I've led those organizations and help bring those when I run internship programs. I have run organizations and kind of mold both approaches to doing it is you need to build consensus within those two forces, if you will. One of that is to have a strong sense of empathy and respect for what has been there before. One of the first things that is always a concern is people who have been there doing it, nobody wants to do a bad job. Everybody is doing the best that they can with the information that they have, with the knowledge that they have, right?


So now, when you come onto the floor and have those conversations, be respectful of the fact that things have been done for a reason. Understand what those reasons are before you provide alternative approaches to solving that problem, right? I'm not saying that alternative problems are not going to help solve them. It might very well change the way that things are being done because we all know that it's a sunk-cost effect. We have sunk, we have expanded, so much of our investment in a particular technology and a particular way of doing it. It's human nature to not change that. "Hey, I'm doing this. It's worked really well for me in the past. I know how to get this technology to work for me." Right?


It's human nature. Understand that. So I kind of work from both angles of it to kind of help them see there's value in it. And then you kind of... One of the problems in it is to kind of force function that activity and end up saying, "We need to get this problem solved for this particular reason and by this particular amount of time based on these cost structures that we have today."


And so, not to go off and build this fancy thing, in terms of polishing it until it looks nice and good. But we need to solve this problem. So there are certain things that have mechanisms that are done that allows us to get to the point where we need to be.

George Jagodzinski (10:55):

I love all that. The empathy part really resonates with me because I know, even personally, I've gone through the journey of, when I was younger in my career as a software engineer, I'd do in, and you go through that phase of, "What is all this garbage? Let's throw it out, and let's write it over." And then, as I've progressed in myself and us as a company, we've done various, countless audits and assessments of what's been done. I've almost turned it into a more positive-looking game of, "Hey, all this stuff was done for a reason. I bet we could reverse engineer what those reasons were as we look at this," right?


They were under a time constriction, a resource restriction, whatever it might have been, it was... I really like the fact where you said, "No one did that because they wanted to do a bad job. It's just the nature of the constraints that you're within," right?

Kishore Aradhya (11:42):

And also, another thing I try to have them see is, what about the future version of you 10 years from now? You're talking to a person, you yourself will come back and say, "Maybe I shouldn't have done that." Hindsight is always easy. Always come back and say, "I could have run it differently." I'm sure you'll feel the same way. I felt the same way when I came fresh out of the journey. And there were times like, "Well, maybe I shouldn't have done that. Maybe there was a better way to do that."


So we're all human. We're all trying to figure out what is the best thing to do.

Speaker 3 (12:10):

Evolving Industry is brought to you by Intevity. We bring order to chaos wherever people, process, and technology converge. Our culture drives our solutions, and we are solution-obsessed. We see every challenge as an opportunity, every partner as a collaborator, and every project as a chance to share our values and commitment to excellence. Give us a shout. We'd love to hear your challenges and turn them into opportunities. Find out more at intevity.com. Now, back to the show.

George Jagodzinski (12:40):

You know, you talked about the generational shift, but I'm always very drawn to the big organizational shifts. I think you've been at quite a few organizations that they're shifting possibly from B2B to more B2C or a balance of figuring out how to more attack both of those channels at the same time in smarter ways.


I just think about Stanley Black & Decker. What I love about that company is 180 years ago, they started just making bolts and hinges for doors, right? And now, I'm assuming, are inundated with data and trying to figure out what they're doing with data. And I'm curious as to your perspective on how do you make sense of that? How do you prioritize? How do you play into that shift from a... Is it a B2B or B2C, or is it just how do you engage with your B2B in more interesting ways?

Kishore Aradhya (13:30):

Yeah. That's interesting that you bring up Stanley as one of the more recent things. I also worked at organizations like Staples, which was both B2B and B2C kind of thing. And similar to that at Bose, their headphones is a great product. And again, that is a B2C side of the equation, as well. But there's also some elements when you work with the automotive world where both systems were in there, so you have a B2B there, right?


So it's an interesting thing when you frame it in terms of B2C or B2B. I think that world is kind of converging to the point that it's like data is getting created in an operational, in a transactional way. It's almost like the digital, I hate to use this word, digital exhaust, but it's one of those things that's been there. It's the existence of a business creates this data inherently.


So the question then becomes for data from a data engineering or from a CDL perspective, or whoever it is that I've engaged with in the past. One of the challenges is how do you detect the signal from the noise? So part of the challenge is what are some of the underlying approaches you take to that? Have you used relational systems in the past and continue to do that? But also, the graph-based approaches to doing it, the knowledge-graph approach to doing it. These are some of the things we were trying to explore. To bring me back to the point, it's kind of back and forth. As you said, we have thousands of products, and we have actually quite a few, I would say close to 20, 30, 40, 50, brands, if you will, under the MOA over the years acquired. So you go to all these different, especially the supply chain things that we went through recently, and it's still kind of going through with that, is how are you able to map out these different parts?


For example, you have to build up materials, you have so many different things to it. How do you reason around it? How do you build systems around it so you can ask the right questions with that or to that, so you can get the information that you need. You have to go invest in the right kind of areas, or build the right kind of products so you can sell it. Again, you're selling in over 190 countries, so you have to have a sense of what is being done, right?


And also, the folks who are using these tools or using these mechanisms are very different. They may not be digitally savvy as you would expect them to be. But they know a good product when they see it. So then, the question then becomes how do you incorporate these signals into these machines? A lot of them are moving into electric systems, right, electrical cars, etc.


So those are all, can clear these signals, if you will. And how do you pull those systems out of it? Now it's a manufacturing unit, right? You've got a lot of those things going through with it, so you've got to create these manufacturing digital twins, if you will, so you can then start to spit out, "Okay. How can we make things better? What are all the signals we can get from it? And what other kinds of things attract from that?"


So without digging, going too much into details, those are the contours of some of the things that we kind of think through.

George Jagodzinski (16:44):

That's great. It's really interesting. I'd love to double-click a little bit on the knowledge graphs. Some of the audience might not have any idea what that even is. I'd love to hear it. How are you seeing them being leveraged? What are they, and how are you seeing them being leveraged?

Kishore Aradhya (16:59):

Yeah. When you think about it, right? This is an interesting thing as I'm kind of teaching one of those classes at Northeastern. I was going back and kind of mapping out how we ended up in the relational world that we are in today, right? And back in the late 50s or so, if I'm not mistaken, there was a group that had come together, I think it was Cortisol. I don't remember the exact number. But what they put together was, they put together three or four of those things. And the graph aspect to that, the ability to connect different things, the relationship side of it, was part of that original, shall we say, the thinking behind it.


But the problem was, the systems, the computational systems, the things that were there in those days, were not there for it to... It was turning out to be extremely, shall we say, challenging to computationally provide the query mechanism to get the relationships in the way that we do today.


So the relational algebra, that's what really took off. It's basically two math sides to it. There is the relational, that knowledge side of it. And then, they have all those set up in it, and then there's the graph theory side of it, right? So basically, all the computational elements and all the thinking card knowledge folks came together, put all the pieces together. This kind of really optimized and took off, right, with the Oracle, with the IBM, all those things got really focused on that to help the capability there.


So when that happened, even though in more recent times, in the last 20 plus years, you're seeing data that getting more focused on the relational side of it, the networks of it, if you will, whether it's the social networks that kind of really created this entire area. I'll put it this way, you've got the noise skill, even though I hate the term noise skill, but anyway, all those things kind of came through before.


So we started to realize that not only is the entities themselves critical, but also the relationships between entities are equally, if not more, critical and needed. So you've got the Linkedins of the world, we use it a lot. So at the end of the day, those things are all relational in nature.


And in a similar nature, it doesn't have to be social, right? It could be supply chain, as we just talking about earlier. Or, it could be the parts, for example. What are the parts that go into making this particular tool? That particular little part, or screw, the size of the screw, could be used in thousands of other products, right? And then, you start building relationships around those things.


So you could start to see things around you, and you see a lot of those things. And especially, nowadays, for example, a whole bunch of graph... I don't want to mention one or the other and not talk about the others, but there's a whole bunch of areas coming together, and with the, generally, with the LMs coming together, you're going to see a lot of those intersections of how LMs can build, construct, knowledge graphs, can help build those anthologies, can help build those values within that thing.


That, to me, I think, is a very interesting space that's going to grow further as we, as you can because people are getting challenged with the rag models that are currently being built with, and that's the typical standard thing, right? Whereas here, with more of a static, more anthology driven thing, then you get real data faster. And more, shall we say, accurate data.

George Jagodzinski (20:29):

Yeah. I love it. I've always been a big fan of link analysis and network digitalization. I was doing that a while ago. But for the soul reason, it becomes so easy, so much easier, to discover things in it that you hadn't been thinking about. As an example for the audience here is, in a network of people, you might have one person who is the connector, right? That one person that is the one person between multiple groups of people. Or, you might have some person who is unique in that they're completely off to the edge from everyone else. And you apply that to parts and you can now, all of a sudden, see where that choke point is in your supply chain, right? Or you can see where the opportunity is.


And then, when you start layering over AI on top of there, then discover what those choke points are and find different ways. It's just a more natural way of discovering these things, right?

Kishore Aradhya (21:24):

It absolutely is. When you apply the reasoning engine… there's now no motor reasoning engine, it's flawed, but still a reasoning engine. When you apply that against this anthology map, if you will, against this, you're going to see some interesting value proposition. I think we're still in the early days of that. But I see a lot of organizations, a lot of startups, a lot of those things beginning to focus more on that.


So I think we're going to see, as you said, we as humans naturally of things as relationships. We do not think of it as... Things don't live by itself. Nothing kind of lives on its own. Everything is connected to everything else, right? And so, the real value is at the edges, is at the connections. It's less to do... I'm not saying that it's not important mechanisms, those are important, obviously. But I think we solved the problem, in the sense that we are many systems, many optimizations around to help to get that.


But how do you get the relationships, all the different pieces together? And also, how do you think about it in terms of applying to other areas? People get too focused on, “Oh, the Linkedins or the Facebooks of the world…” Yes, they are important, but the underlying aspects to that can be applied in so many other areas, in health-related areas. Think about that. All the different billing systems, mundane things like that, right, financial stuff.


So think about all the different touch points through all these different things. So I think we're at the very early stages of that. But I get excited talking about that. That's interesting.

George Jagodzinski (22:47):

Oh, same with me. I mean, as the wise philosopher-musician said, "The hip bone is connected to the thigh bone," right? So everything is connected.

Kishore Aradhya (22:54):

That's right, that's right, yes.

George Jagodzinski (22:56):

So stepping out of graphs, with your many years of experience and your interactions with the educational facilities and the students right now, what are you most excited about looking forward?

Kishore Aradhya (23:10):

I think, you know, they're entering an area where there's going to be extremely challenging for a lot of the folks, right? It's not like when we were entering the field, where you learned... I'm just making it up here, oversimplifying it. But you're really good in one programming language, whether it is in those days, C, C++, whatever it was. Then you could kind of use that to kind of fork an entry into different aspects of the areas. And obviously, you needed to know data structure, the standard CS stuff, right? I'm not saying that's not important.


But the thing that really kind of drive home the point was in one of these things. But nowadays, your knowledge should be so varied. It needs to be a lot broader than the need used to be because there's so many touch points in which things are happening. And the goal nowadays, honestly, I think, has less to do with going very deep into one particular area at the cost because the opportunity cost, think about that. Your goal... You need to be able to have a pretty good intuition on different aspects of it, whether it's software engineering or whatever.


Now you have all this in LP. You have all these different areas. So when you think about all these areas, you really need to be able to reason at different levels, across, horizontally, while you kind of pick one or two areas, the focus areas, if you will. And then, you kind of go deep into those things. But you cannot go all over the place. And I find myself having these conversations with the students. Like, "What about this? What about this? I want to do this."


"Yes, you can go after doing so many different things, but then you're going to lose sight of what you are focusing on." So I think focus becomes a huge, I think, it's a huge challenge, honestly.

George Jagodzinski (25:06):

Yeah, yeah. I'm seeing more and more the rise, this is the age of the generalist, right? Or, we talk about our M-shaped or our T-shaped resources, where you go broad and then deep into one or two. That's very much in alignment with what we've been building out, what we're seeing out there, as well. I'm curious, as someone who owns and runs a business that's in this space, how can I best be leveraging this new round of talent that's coming out?

Kishore Aradhya (25:35):

Yeah. I think it's very interesting because when you think about this. We are, obviously, I think, have been much used inflection point, right? Just like the way when the browsers and things in the late 90s came on board. People didn't really realize what it could do, right? They were like, "Oh, just throw up a webpage. What does it do? Really, it's not going to change our lives too much," right?


And then, we saw the evolution of that to where are today where it's like, "Of course, they're going to use it," right? We are at an inflection point, but this is happening at such an accelerated pace that it is almost in all of humanity. I don't think we have seen this kind of disruption that is going to come our way.


The question then becomes, given that fact that these general-purpose technologies... I use the word general purpose trying to put this in general. That could be used in so many facets of a business. You're talking about a business. Let's say you have a business. You have a business workflow. You have a mechanism in which you want creating value from one end to the other end, delivering value to your customer. So all along that line, you see a lot of different opportunities for automation, for example, optimization and so forth. You see all these paths along the way.


I think the organization that is able to map that to the underlying technology and not just... When I use the word technology, I use it in a general term. People, process, technology, obviously, are the biggest part of that. To get the right people to look at it in those approaches helps you optimize it faster, helps to deliver value faster, that, to me, is what is going to differentiate success from failure. If your organization needs to be successful, you really need to get down to that level of it.


And leverage, leverage, leverage, experiment, and try it out in different areas.

George Jagodzinski (27:27):

Yeah, and be really nimble as you do it. I was thinking back recently to the days of two, three-year implementations of platforms, and I think those days are completely-

Kishore Aradhya (27:38):

Long gone, are long gone.

George Jagodzinski (27:38):

... behind us, right?

Kishore Aradhya (27:39):

Yeah, absolutely, absolutely. Anything that takes you more than a few weeks, you need to question it, honestly. I mean, it's strange when you say it, but I understand larger organizations, it's not going to change overnight, drastically. But, interestingly enough, large, complex organizations, even if you can make small tweaks, things you and I were talking about earlier, a small tweak here, a small tweak there, could then translate to a larger impetus, down the thing.


So I think that's the key there, is to understand where you're going to provide the biggest value. And don't think in terms of months. Forget about months. It takes several weeks to get things done.

George Jagodzinski (28:21):

Yeah. I mean, even just looking at physics, you can move very large things with just lots of little things. It just makes a lot of sense.


Kishore, thanks so much for being here. I really enjoyed this. I always like to finish with a fun question which is, in life, in your career, anywhere, what's the best advice you've ever received?

Kishore Aradhya (28:41):

Curiosity. Be curious about everything. Don't accept things for what they are. Always question. Always be curious. And don't ever be afraid of asking questions because nobody has the complete answers to everything. Just be curious and ask questions. Don't worry about looking dumb or anything like that because we all are. We all were at some point.

George Jagodzinski (29:09):

Well, as someone with the name of George, and a slew of Curious George books behind me, I can get 100% behind that advice. Kishore, thank you very much again. This was great.

George Jagodzinski (29:19):

Thanks for listening to Evolving Industry. For more, subscribe and follow us on your favorite podcast platform, and pretty please, drop us a review. We'd really appreciate it. If you're watching or listening on YouTube, hit that subscribe button and smash the bell button for notifications. If you know someone who's pushing the limits to evolve their business, reach out to the show at evolvingindustry@intevity.com. Reach out to me, George Jagodzinski, on LinkedIn. I love speaking with people getting the hard work done. The business environment's always changing, and you're either keeping up or going extinct. We'll catch you next time, and until then, keep evolving.