Podcast with Jaya Part 3
SPEAKERS : Jaya, Liew Wei Da Andrew
Show Outline
Section 1: About the show
Section 2: About the guest
Section 3: Jaya’s tips for Csuite’s digital transformation
Section 4: 2 Schools of Thought
Section 5: Stories of deploying AI in digital transformation
Summary:
If companies need to enable digitaltransformation, they should focus on the identifying areas where value can becreated, to the linkages between how technology and the rest of the businesscomplement. Jaya believed companies that adopt a decentralized approach first followed by developing a centralized team. This means bringing some of these experts, andeventually, these experts will form a centralized stack. It will be going from a decentralized to centralize over time, so that companies can have economies of scale, and yet also have accelerated skills transfer to the juniors over time. Along the way, Jaya shared some stories of digital transformation.
Liew Wei Da Andrew 00:18
Hi, everyone. Welcome to the AI of mankind show where I shareanything interesting about mankind. I'm your host for the season. My name isAndrew Liew. I've worked across four continents and 12 international cities. Also, I work in tech startups across a range of roles from selling products, making customer happies, figuring out fundraising, making finance tick, building teams, and developing sticky products. Apart from building startups, I've also worked in Fortune 500 companies. As a chief data scientist or technologies, or people leader. You can call me Jack of all trades or master of learning. I hope to make this podcast show a great learning experience for us.And each season there is a series of interesting things, where invite guests to share their views about their life and interests. Now, let the show begin.
Liew Wei Da Andrew 01:27
Hi, everyone, thank you for coming to the show. kindly allow me tointroduce my guest for today. Jaya or Dr. Jayarethanam Pillai. Jaya is the chief innovation and data scienceofficer. He has run his own technology startup company and consultancy firmoffering regional experience in providing analytical and data science work for companies and government agency in the likes of United Nation, Asia Development Bank. You know. And also particularly specializing in developing strategicdecisions using economics, data analytics and artificial intelligence. Jaya has done over 20 analytics related projects and widely published papers in economic policy, digital transformation and entrepreneurship while working and living in 11 international cities. He has expert level advisory experience in deploying end to end data engineering and data systems. In addition, he has obtained his PhD in Economics from the Australian National University. Let's put our hands to welcome Jaya.
Jaya 02:51
Thank you very much, Andrew, for having me part of your photos.It's such a pleasant surprise. I'm pretty good.
Liew Wei Da Andrew 02:58
Okay, so the purpose of this podcast is to really understand topicsin AI, artificial intelligence, understand digital transformation, and alsounderstand about your story. So, yeah, okay.
Liew Wei Da Andrew 03:10
So you mentioned about, you know, the fact that digitaltransformation can be affordable, people can save money on digitaltransformation, there is definitely a need for business reinvention. It's not a choice but it's a must, there will not be a new return to norm.
Liew Wei Da Andrew 03:28
So you pitch this case, to C suite or CEO and he will come back toyou say Jaya, I agree come, let's do it. So my question to you here is what doleaders or C suite executives need to take note when they do digital transformation?
Jaya 03:46
The pandemic dramatically kind of accelerated the technologyadoption across major major industry. So in my mind, there was a particularstudy, which surveyed CEOs in the US companies talk to you 77% reported that they to these companies to speed up digital transformation plans. Why? Because the whole entire decision making has completely crumbled during the periods ofthe pandemic, right. So like Microsoft CEO, kind of a highlighted, right that we have, we have seen two years worth of digital transformation in two months. You see two years to two months. So if companies want to make change, they can make change on the fact that usually takes six years to 10 years to make major transformations. What is interesting is that this pandemic, a simple virus has completely changed. Executives should actually concentrate and how their roles have been changes. Like, for instance, I have several in my mind, let me just say one or two of those instances, so chief technological officer, since I working on digital transformation and digital strategy is that in the past CTOs were the resident experts. His opportunities and limitations were presented bynew technologies. But now CTOs go beyond that. They are called upon to lead company wide digital transformation. That's a lot of work, right? They not only have to formulate new strategies, which means when I say new strategy company decision making, but more new digital strategies to incorporate in decision making, right, they have to be achieved, played in the decision makingstrategy, right, identifying areas where value can be created, to the linkages between how technology and the rest of the business complement, because it's becoming central role, and how that can motivate an ally fellow employees. Remember, we just don't look at the top management, we have to go back to the ground. That's where the actual fellow employees are, how are they? Are they happy to embrace such changes? You just don't want to say tomorrow onward thateveryone shift to using python that everyone must shift from using Microsoft to say, Linux. No, it doesn't happen like that. Right? It's a gradual embracing of new initiatives and technologies. CTOs are very crucial in inviting such interests. How do they cater? And how do they develop that skill sets so that they are paid for that job? So they have to know how to do that?
Liew Wei Da Andrew 06:24
Yeah, you talk about this is this is really interesting about CIOCTO. So there's always two schools of thought. One is we centralize everything,like centralize the data set system, centralize the software, or either we go for extremely everything open source, right. And their argument is economy ofscale, or manage costs. But then the challenge is that the CHRO will say, hey, we couldn't find the talent to do these vehicles is in scarcity, for specific technology for a specific approach. Then, the other extreme school of thought is that the CTO CIOs say, okay, these days, talent is more important, like we are aligned with the CHRO, we will get the best and the brightest to be BYOD,bring your own stack. That means if you're a Python guy, bring your own Python, if you're a STATA guy, bring your own STATA. So bring wherever you can. And as long as it's within the ecosystem as allowed, we will empower you. So what is your viewpoint between these two schools of thought, and where do you stand?
Jaya 07:30
I say it's more on the decentralized situation. And I stronglybelieve in that is because just bringing a top guy to sit on boards, or sit onmain positions, doesn't transform into interesting ideas, or interesting processes or interesting developments. What we need is a balance. We have theright guy freshly out of graduate school, and a guy who is able to actuallykind of like allow the slow development of more in training in person andcatering those talents to the needs of optimization guidance. So we need tobalance on both yes, we need good, talent. But that doesn't mean that we should jump and grab the best talent out of MIT or NUS.
Liew Wei Da Andrew 08:15
So and so your hybrid is saying that, okay, we will bring this decentralized:the bringing some of these experts, and eventually, these experts will form acentralized stack. It will be going from a decentralized to centralize over time, so that you can have economies of scale, and yet you have accelerated skills transfer to the juniors over time, am I right to say that?
Jaya 08:40
sure, yes Job rotation is a very key area. Because what I have learned andexperienced through my own studies, and at the same time application in thereal world is that no one particular leader or management role, know it all. That's the reason why we have so many variety in terms of how digital transformation is looked at, right? You can be some guy who is actually just looking after documents, but that the guy who knows what these documents are leading to, he has a huge influx of knowledge about the search documentation. Putting him in a job rotation allows him to impart such skill sets such understanding towards other different departments, right. So, job rotating is a crucial factor when we talk about this hybrid model, right? The ability to understand comprehend all these skill sets and then go on to lead the company. But at the same time turn back and say we need to develop skill sets through job rotation. Responsibilities has to be shared.Responsibilities of different expectations have to be placed upon. These will inculcate this hybrid model of acceptance, right? But it's not easy because embracing doesn't take one person, you need the whole company, you need two hands to play. Right. So that's how I look at it. The 2 schools of thought, some might say centralizing. Increasingly with this remote style of work, weneed more decentralization, people are more happy when they are given responsibilities. Because they want to feel important. They don't want to be one of those who sits in the cubicle, crunches the number or writes the document and sends an email and doesn't is not even looked into right, I feel more responsibility allows people to actually enjoy their work a lot more.
Liew Wei Da Andrew 10:36
I see. So coming back, you just mentioned about you know, let's sayC suite needs to take note that there's a need for digital transformation, weshould start with a decentralized approach, hiring mid career guys or experts, and then groom them to become a centralized approach to build and cement economy of skills, enabling the job rotation, you know. And okay, so coming back, let's say, you know, a C suite or CTO,CHRO, come in and tell you, so Jaya, where do you see AI in the process ofdigital transformation?
Jaya 11:16
One, if you look at digital transformation, right? AI is usingalgorithms, so digital transformation is the big block. But to get to there, weneed to use the tools to bring it up to transform digitally, any processes. So AI is using algorithms to create or adjust processes, or programs to takeadvantage into the insights. When I say customer data, that's the insight we need. So by actually digitizing those processes, when I say digitize using machine learning models, AI algorithms, these are being digitized, whichactually mimics the human intelligence to identify or react to behaviors andevents. When that is happening, we have a so called transformation. Right? Sodigital transformation, as it is doesn't just happen, it's actually a long process.
Jaya 12:15
It's a huge block, which we just say AI, neural network machinelearning, you know, modeling to create or modify customer experiences and culture,business processes, and tasks meet customers changing needs in the market, because that's the most dynamic customers, that's the base of any company, right? When I say customers, as I say, this is not only the household massmarket, you're looking at suppliers, upstream downstream governments or customers. So you need to have to look at all aspects of customers dynamic behavior. AI, helps companies to innovate, become more flexible, and adapt. If you have no correct data, or data, which is not organized, let me say is all over the place, if you have a hard time trying to develop this whole digital transformation platform, right, you need to have all this properly catered for and looked after. Because these are crucial. These data, allows how your transformation how your toolscan help you to make it better, right? You might be better, but it can actually make it more efficient. Right?
Jaya 13:25
So the most important is to, there is no way to have artficialintelligence without having a clear defined data strategy. Right? There is nopoint in seriously talking about AI, if we do not have data organized properly. And companies which are staying ahead of the curve is those who are able toorganize, collect this data and protect it securely using it for their, for the purpose of actually getting the best out of their business models. That's what it's crucial.
Liew Wei Da Andrew 13:59
So you're just now saying, Am I right to say that? Where is AI inthe process of digital transformation? you mentioned about, you know, figureout the business case or the business model, and then going back to figure out, you know, what are the data that we need? And then going back to the organization with it information technology to figure out where's the data right and how do you clean the data, structure it and then running applications of AI and machine learning.
Liew Wei Da Andrew 14:30
So coming back to that, like, what are the applications or AI andmachine learning that you have done in the course of your career, give us two or three examples.
Jaya 14:43
What kind of processes I have used?
Jaya 14:46
I gave you an example when I was working for this engineeringcompany, where I used develop a forecasting model on a for the shipping companywhere I kind of was able to predict a three month pricing strategy But I was able to collect the vast amount of for what it called historical data andmanipulate it through using economic metrics, database for actually identifying the ups and downs and peaks or where shipping lines are affected through different environmental issues. Right. That's one, two, how logistics and supply isn't affected due to many policy related. So it's a verymultifunctional database. It's a multifunctional modeling and forecasting, so that such companies which offer services are able to actually price the packages ina more what you called more competitive.
Liew Wei Da Andrew 15:50
So at a high level at a high level that you mentioned, it was tohelp that client figure out what is the future price, so and so in the courseof figuring out what is the price, you will have to figure out? What drives the price, right? What is the input that will drive the output?
Liew Wei Da Andrew 16:09
But there are so many different open source algorithm, is there anyalgorithm or any principle, just high level that you actually use?
Jaya 16:21
Yeah, I would, I would say, as opposed to actually looking at ageneralized model, because such generalized packages never really help theincentivised people for this, these particular companies are looking for. They want something which actually caters for their profit margin, right? They wantbecause you have to cater such models to the kind of data and activities they do. So when, for instance, I could bring in the informal modeling, right? And calculate a pattern where we could see what are called distribution, distributive trade, right? But that's not the aim, right? If he can't do that, with companies, specifically, we have to develop some kind of database where they can retrieve information, like, for instance, how, what is that time period, a single hub to double hub transformation of a ship? How quickly can they actually transform, right? That's one, how were the patterns in the different shipyards yards where these ships are being transformed is affectinglabor force, the existing labor force. The kind of training they have the kind of and of course, you have to look at things like you know, what are the what is the rate at which such yards are able to churn out the new ships, the rate at which they are able to put what you call: the materials in use, so and whether they have the accessing materials for and answering and quickening theprocess of changing the single hub or double hub ships, right? So at that rate at which all the shipping lines, he is able to actually go past their trading routes. So there's, that's one of the another area where I could say, that's one kind of model where we build a billion prices.
Jaya 18:12
Another thing which I can suggest would be the transfer pricingscenarios. That's one of the big headaches we have now. Now we're also lookingat carbon emission models, right? How such models are being actually inculcated into AI and machine learning to actually enhance the understanding of who pays for carbon emissions. Yeah, that's very crucial, right? That's very important now, because we want to know, which companies can actually reduce the carbonemissions, right? In this way, they don't have to pay a penalty, those who have made more, they have to pay a certain penalty, right? At times by their population size, their profit. That's how they look at right. Like for the company, which I worked as a project engineer, as a document engineer, one of the things which was important was when I took over the position is to actually collect as much historical data for projects itself. And using that to actually formulate some form of modeling to see how errors are different. What were their errors, and how those errors could be squeezed out using a forecasting model, error free model. So using an economic metrics technique, which isbasically the what he called the Bayesian rule, or what we call the Bush Pagan test, you can use those kinds of testing to actually minimize and find out the probability of these errors along the line. Right. So but the the key fact is, data has to be organized in a very clear way, very, like very clearly placed so that it's easily retrievable, and also at the same time you store it and youexplore it. Production has real time operation and extraction. Using that we can actually predict and optimize it. So that newer projects, which might be very similar to what we are looking from historical aspect of things can actually eradicate the delay. We can straight go from having documented we know this project, this is how these models will work, go ahead, push it down theproduction line, right. So having this data nicely organized and clearly stored allows AI and modeling to actually quicken the efficiency
Liew Wei Da Andrew 20:37
Thanks for showing me that two use cases. One was the pricing ofthe different haul information and and then you're talking about various othermodels that you try on carbon pricing and price elasticity.
Liew Wei Da Andrew 20:51
Hey guys, thanks for listening to this podcast. If this is thefirst time you are tuning in, remember to subscribe to this show. If you havesubscribed to this show and love this episode, please share it with your friends. family and acquaintances. See you later and see you soon.
SUMMARY KEYWORDS digital transformation, models, companies, data, ai, processes,jaya, ctos, called, pandemic, develop, customers, technology, decentralized,talent, role, experts, important, projects, centralize