#337 Driving Innovation with Data: Ashis Parajuli on AI and Open Source

[00:00:00]

Mehmet: Hello and welcome back to a new episode of the CTO Show with Mehmet. Today I'm very pleased joining me from Canada, Ashis Parajuli. Ashis, thank you very much for joining me today. The way I love to do it is I keep it to my guests to introduce themselves because I [00:01:00] have a theory no one can introduce themselves better than themselves.

Mehmet: So the floor is yours, Ashis.

Ashis: Yep. Thank you. It's my pleasure to see you here, Mehmet, and, uh, uh, it's glad to, it's glad to present and glad to talk in the CTO show. It's an interesting show that I'm getting part of it. Uh, yep. Thank you. And this, this is Ashis. I'm founder and CEO at DLytica, which is AI, um, data science, machine learning and data and, um, engine data engineering analytics company.

Ashis: Like we do the, we do provide consulting and we do also have the. Big data platform. And, uh, basically I'm working on the innovating, building the innovation of big data and AI world.

Mehmet: Great. That's great, Ashish. Maybe a traditional question a little bit, but what inspired you to be in this field and, you know, Uh, to focus on, on these, uh, nowadays very important aspects, [00:02:00] data, cloud, AI.

Ashis: So, uh, definitely this, this AI thing is getting, uh, kind of boom right now. So everybody, you see, you go everywhere, every organization wants to have AI in their business. They will, they want to leverage AI to boost their, uh, AI. Uh, boost their basically, uh, business, uh, with the AI, boost their, uh, productivity with the AI or boost their day to day, day to day cycle with the AI.

Ashis: So basically. This is interesting and this is so exciting that every organizer feeling that they must be, they must be, um, catching up with this game. So I was in the very forefront of this, uh, game, uh, very long time ago. So it's been around, uh, 15 years I'm in this field, um, ground up, like I was working and building my, my career itself into AI and data science, you know, so I, basically I was very much directed into How I can do the [00:03:00] digital transformation through AI and data things.

Ashis: Basically the goal is To, uh, to collect the data from, uh, different systems and applications within the, within the organizing itself and, or maybe third party data, data sets, wherever the customers are, uh, customers are, um, like engaging into, you know, so that way we get to know about our customers, we get to know about how our business is operating, get to know about how, what kind of activities is running through the organized and that way we can leverage the AI, the data to, you know, Take the decision, you know, so the interesting thing is data driven decision is something that excites me and, and that, that is what I try to leverage as well in my day to day activities.

Ashis: Uh, you know, so that's, that's what excite me and that is. where I am and I'm trying to help the business or the people that they're leading to this, this journey.

Mehmet: Perfect. You know, like as I was telling you, Ashid, like this is, uh, the hot [00:04:00] topic today and everyone trying to leverage data. Now you mentioned something that you are driven by, you know, helping customer, building solutions, which are mainly, we call them data driven because, uh, data is, is the new oil and actually worse than more than oil now.

Mehmet: But I'm sure that you see some challenges across both maybe even startups and enterprise. When they want to implement data driven solutions today. Um, so tell me about these challenges, like what kind of challenges you see in the field and how do you help them navigate these challenges?

Ashis: Yeah, that's a, that's a good question.

Ashis: Uh, that reminds me of, uh, like the situation where I was working with some of the clients and they were asking, Uh, questions about, uh, Hey, I have, uh, I have this idea of this idea of building an AI [00:05:00] solution. I have an idea of the link in this AI use case, but I don't have, uh, the idea how to implement it.

Ashis: I don't know. I don't have, I don't have the resource within the organization or I don't have the skill sets, um, in my organization or, uh, or am I mature enough to, uh, implement this? So this kind of question, a lot of questions comes into, uh, my, my area when, when I start talking to customers, but definitely the, like, like I said, these, the question, whenever the questions comes in, they, these questions are like carrying the, the challenges.

Ashis: So the challenge is definitely is the, uh, first thing I like I shared is a skillset or the ideas that. That organization would like to have, yes, they do have the data. Um, right today I was talking with one of the customer, you know, like, uh, it was a banking and they have like a terabytes of data running in the organization.

Ashis: But right now they're, uh, business is asking businesses. [00:06:00] Not asking much question because they don't expect to, because they are not, they're not yet familiar to and, uh, uh, the IT team or the, the data team, like the, the, their IT team was not, uh, was not delivering something that maybe businesses are asking and they were not delivering.

Ashis: It's something like data was running in silos and like the idea is the data was not being leveraged. So. Like I said, the why why it's why it was happening is because of the skill sets, definitely they don't know how to bring the thing, the data from different platforms, the systems in one place and build a let's say customer 360.

Ashis: Uh, let's say for example, we want to know, uh, where the customer is, which demographics the customer is from, or what kind of the, uh, activities our customers are doing. Let's say if it's a bank, what kind of transactions, uh, the customers is, is doing. So how often he, he or she is using the, uh, the products or [00:07:00] banking services, let's say the saving account or credit account or credit cards or loan, how often the customers is using it.

Ashis: Or let's say how often the, or let's say if it's a telecommunication, then, um, like, like understanding, uh, like where the customer is calling, like how often the customer is calling to abroad, or maybe something the customers might be, uh, might not know about roaming service and he's calling international and getting more charged and, or maybe some of the transactions, like maybe activities of the customers, they, they interact with our advertisement in, let's say, if, if a telecommunication is.

Ashis: Promoting a service or internet service offer and the customer is interacting through the, um, through the ads. So now the interesting, what I'm, what I'm going through is, is about, um, the, like, if you go with the customer journey from initial advertisement or initial customer onboarding. Uh, to, uh, the customer, [00:08:00] day to day customer activities.

Ashis: There the data are, there, is there is data being generated? So it's up to use how, how you leverage, let's say for example, if a customer. Uh, reacts in your ad or maybe customer so interested in your ad that is first data that you receive in your organization. Okay. So let's say, let's say now you want to know which product or service the customer might be interested in or proposing the right product in the right time is maybe the first data driven decision what you have to do.

Ashis: But right now that the challenge, major challenge that the organization is having is proper. They don't have the proper platform to do it first thing. And definitely the skill sets. And the third thing is the vision towards the data. What can be done with the help of data? I think these are the major three, uh, three areas where I can see the major challenges.

Ashis: So when they are clear about these things, um, definitely they should start leveraging the data with the help of right skill sets. [00:09:00] With the by using right technology tools Uh, maybe process people process and technology is something what they need to have in the data itself

Mehmet: And this is where you come into the picture.

Mehmet: I believe Ashish, right?

Ashis: Exactly

Mehmet: Now you mentioned something which uh, you know It enlightened a bulb in on top of my mind. So you said they know that they have the data, but the problem is You know, there's too much noise maybe around, uh, what can be done with this data and maybe, you know, they are getting Involved or driven by different things at the same time.

Mehmet: Now, I think the issue and correct me if I'm wrong, Ashish, like data fragmentation is, is the major challenge, as you said, that that there's the scale and you mentioned something about, you know, which I believe is where maybe the CTO should play a role here in leveraging this data into [00:10:00] business perspective, right?

Mehmet: So, so trying to, to understand actually what the business would care about, and then see if we have this. data and then apply the platform over there. Now, why I'm mentioning this because majority of the time I'm surprised that actually sometimes, and have you seen, or are you still seeing this customer?

Mehmet: They don't know that actually they have this data. Like, have you seen something like this? It's like, they're not aware that actually this data exists with, with, with them. Have you seen something similar?

Ashis: That's a, that's a very much good question because, You know, like, uh, in organization, uh, like data on the top of data, it's something, uh, what we call the data glossary, you know, something, uh, definitely people in the organization, most of the organization, like I said earlier, they don't even know what data they have, like, right, like you said, like, it's a major, major challenge.

Ashis: They don't know how it can be leveraged [00:11:00] or what can be done with this, with this data available organization. For instance, um, let's say for instance in an, in telecommunication, um, if they have, uh, let's say these, the, the, the CDR files, the with the help of CDR files, you get to know, okay, where, where the customers are calling, how many, how much time they're spending in a, in a, in a data.

Ashis: They're in a, in a call and if their call is being interrupted in between or not. So sometime, uh, now business, if business knows, okay, so if I go back to networking systems that I will be able to get, um, okay, when the disconnection happens and what was the network activity while the network disconnection happens.

Ashis: Okay, so I'm going more technical now, but just to go a little bit deep dive, people get it. So what I mean to say here is to find out, to find out, uh, when the customer is, uh, is getting disconnected while they're [00:12:00] calling the reason they, so why I'm talking, taking this problem is because this is one of the major reason why, why the customer is being churned in the organization, why, why, why the customer is being leaving the, the organization.

Ashis: Uh, the, the telecommunication and going to all the example, right? So let's say in that case, like let's say the business would like to understand, um, maybe in traditional way, what we do is just call the customer and try to understand why, why they were moving all the organization. And this might say, um, because of this, because of that, and then so on.

Ashis: But the underlying. thing, what I'm trying to say here is we have the data and the business itself knows, uh, business self knows, uh, the reason they don't even need to ask the customer. They knows, they knows themselves, you know? So what I mean to say is, let's say if you go to the, if they go to the CDR files, it will go to networking systems and they get to know, okay, why, why the customer was just when the customer disconnected, why they were disconnected and what was the reason?

Ashis: Maybe is it because of the server failures or a visit visit [00:13:00] because of the, uh, the, Uh, poor networking connectivity or visit because of something. So maybe we get to know. So some, let's say if the business knows themselves, or let's say CTOs, uh, would like to, um, would like to explore this kind of, uh, to solve the problem, then they would explore this kind of solutions and find out if they have the data.

Ashis: So definitely, Um, there are different departments or different teams that they're they're handing those maybe networking department is handling the networking systems like the other operational different departments handling their own operational databases or systems. Now stakeholders gathering is where something that like CTO usually gather and try to understand what kind of data they have.

Ashis: Let's say if they talk to the networking team, they get to know, okay, what really they have. And if they talk to the other operational system database, like the database managers or teams, they get to know what kind of databases they have. And then the CTO is, is really a. It's a, [00:14:00] it's a, it's interesting position where, you know, like, um, that guy should be, uh, as a mediator or, uh, liaison between the business and the real technical teams.

Ashis: So you should be able to translate the business, uh, acronyms or business, business requirement to the technicalities or technical to the business. So that way he gets to collect, okay, what business, what is the business problem and how can I solve and to solve this problem? Where I can find the data. So, so connecting the stakeholders is definitely the CTOs, CTOs, uh, key things that I can say here to, to tackle that challenge.

Mehmet: Absolutely. And yeah, so because also you mentioned something very important and I'm relating this to a little bit now startups. So in big organizations, probably they have collected a lot of data over the years. Maybe they will still have like some missings here and there. And then of course, they will [00:15:00] implement, you know, data collection points, uh, to, to get that data.

Mehmet: And this is why I keep, you know, telling startup founders, like, especially because they, they focus on growth. They focus on customer retention. They, they focus on, you know, a lot of these metrics that people, especially in SAS business, they like. So I said, you need to collect this data early so you can start to take decision early also as well.

Mehmet: So, because if you don't collect data early enough, You are just dealing with hypothesis, right? So you are just saying, ah, maybe the customer is not liking our UI. Maybe the maybe when they are clicking something, there's nothing called maybe, right? You should, it's either yes or no. There's nothing like in between.

Mehmet: So that's, that's very important. Thank you for bringing this, Ashish. Now, on the other side. I know that you talk about a movement that you are a fan of, which is moving from vendor lock in black boxes to open source. [00:16:00] Um, uh, tell me more about this, you know, movement, uh, especially, you know, in the field of big data and, you know, AI and why this transition is important for everyone.

Ashis: I think that's a, that's a, uh, good topic that definitely I love to, uh, help organize and to get out of the bender locking situation to the open world, you know, like the open. Is something, uh, where the, the innovation comes in, you know, like something if it's more branded locked in some of the cases, yes, the, only the sort of organized and let's say the Microsoft or only Oracle, only the, the imagine they have the solution, but some of the cases that, uh, that the, if the innovation to happen, it's, it's an open, open source, you know, so when, when, what I mean to say open source, when I say, when I'm saying open source, it's It's uh, [00:17:00] free to contribute.

Ashis: It's not something free to, it's a free, it's having said that this definitely is, it's being contributed by definitely the, the world, the world is there to solve the problem together. So this is the interesting thing here. So when I say open, Open source. This is, uh, this, this is where the innovation happened when I was already saying that the innovation should start happening when they have the open concept.

Ashis: So let's say an organization, uh, most of the organization that I see in Mehmet is that the, um, that they, they have a platform, but it's a black box to them. So this is a very, really a big challenge. So that's a big, that's a big

Mehmet: claim Ashish. That's a big claim. Tell me more. Why, why it's a black box? Yeah,

Ashis: it's a black box.

Ashis: When I'm saying a black box, like let's say the business wants to do something on the, the black box. The thing is the vendor systems, they, the business, like let's say that, uh, a business [00:18:00] would like, would need to Uh, go back to the, the, the vendor and the vendor who goes to the underlying, uh, whoever is developing, let's say it's a Microsoft product, Microsoft, Microsoft, uh, data platform.

Ashis: Let's say if it's a cloud. Let's say, for example, let me take a cloud era system. Let's say, um, our organization have a big data platform, which is called cloud era. Um, they would like to do, uh, some data analytics and AI. So let's say, let's say that they would like to develop some AI or machine learning use cases, but they would like to add more tools that they would like to have.

Ashis: Or maybe some of the, maybe sometime it depends on organization, organization, they would like to add more data sources, or maybe I'd like to add more other third party different tool, or they would like to more, And more analytical tool or some AI tools, then they need to go back to the, uh, cloud area and the cloud area need to, uh, let's say, um, look back and they will say, okay, this is in my timeline because the, the, because the organization don't know about the platform itself.

Ashis: They do nothing. I mean, they have no [00:19:00] luxury to customize the platform because the business doesn't know. So now they need, they ultimately need to tell to the cloud area and hey, cloud area, I need to do something. And cloud area, what cloud area says. Okay, uh, you can, uh, okay, this is an interesting feature.

Ashis: Let me take it to the my team and I'll get back to you and they say they will take into their timeline and the timeline should be next quarter or next year or maybe within three years. I mean, some of the, some of the features they, they might have in a roadmap and they don't allow you to develop. They don't, they don't build it for you just for you quickly.

Ashis: they might be collecting some more feedback from other customers and then building it. So when I'm saying black box, like the black box is something that you don't know how it's built or that you can customize it. So, uh, this is, this is a situation happening even with the Oracle, there are sort of data, data platform from the Oracle, there are data platform from Hitachi and there are some other organizations that they have their own big data platform.

Ashis: The thing is, It's a black [00:20:00] box. Having said that you don't have control over the way you have to live. You only have luxury to use whatever the tools that they are providing, they're providing and, and the features they are providing. Let's see if you have to add more features or add more, uh, more requirements or more, more use cases.

Ashis: Then sometime you, you might stop with some specific use cases in organizing that you need. Then you have to ask the vendors and the vendors would wait, ask you to wait for a year or so, which is, which is what's really terrible in the, in the cases. And I've seen, you know, I have, I have worked with so many customers in the, uh, Middle East and, uh, Asia region and North America region that they're moving away from the vendor locked in situation.

Ashis: Let's say for example, cloud data Oracle, they are removing cloud data. Oracle and those systems, uh, to the open source world because ultimately the feature wise functionality wise is the same tools. What they are offering is the, the same packaged and more [00:21:00] wrapped version of those solution and they're giving you no luxury to customize, you know, so this is happening, this is happening a lot.

Ashis: And all the other interesting thing is that they are charging so much high subscription cost. It's a yearly cost that you can't get rid of. Imagine that you have a flat, you have a, you have a data analytics initiatives and you start building use cases. You start using your big data platform, big data is really used.

Ashis: And you start, you have to scale up going forward with many use cases that you develop. The worst thing is. The more you scale up, the more you have to increase the server, the more you have to increase the, uh, the resources and so on. Like ideally, they charge, most of the vendors, they charge you per node or per core basis.

Ashis: So having said that, the more compute power that you start increasing, the more prices that you're paying. So most of the organizations are put, like, getting into this trap so that I'm trying to, uh, help, [00:22:00] um, like me, there are a lot of people that know that, that the market right now, that they're supporting open source, supporting the, um, like customization of the open, open source, and then developing them and then giving them the solution so that the total cost of ownership is.

Ashis: Is low enough. So the, if they calculate the total cost of ownership for five years, then the TCO with the other vendor systems versus the TCO with the, uh, customized open source platform is really high and not just a TCO. It's also about the innovation. So their team is free to develop that they, they want to, you know, that the team is free to, uh, innovate with some interesting features, something interesting, uh, implementation or something interesting customization that they can develop.

Ashis: So this is, this is how they foster the. So ideally, you know, the, the reason I'm taking this word is because most of the corporate world are dependent on vendors. So having said that the employees that the technical [00:23:00] team that within the organization are mostly that's a sad part of it, but I was also part of it in the past that the sad part is that the the organization is only dependent on vendor having said that the employees that the technical team within the organization is just assessing the vendors.

Ashis: most of the cases. So, so that's the, that's the worst part of it. You know, that's the sad part of it, but that to break the silo that they definitely should let or encourage the open source. Yes.

Mehmet: Yeah. This is very interesting. Now someone, of course, like I don't have a, a say, you know, for now on this topic, but someone might say, yeah, you know, but these systems, they have been proved.

Mehmet: In, in the field for some quiet time now. Second thing, the support, you know, because, uh, but I got to relate to something else now regarding so support, they will, they will worry. Okay. So if we're going to do or rely on, on an [00:24:00] open source solution, so we're going to provide the support for this. Now, of course, you know, another one might say, Hey, but this is the same thing that happened with operating systems back in the days.

Mehmet: So during the two thousands, you know, when people started to shift Or to move away from the priority operating systems, uh, to more the open source world, like the Red Hats of the world and the, you know, the, the, the Ubuntu of the world, which is now runs probably majority of, of our infrastructure on global level.

Mehmet: This is succeeded. Now, from your perspective, Asish, do you think we can have the same success? And of course that doesn't mean that these systems will not be used and I think you would agree with me There will still some use cases if the customer knows that they don't have much Customization needed and they need something quick out of the box that they can deploy immediately Yeah, probably but yeah to your point if they need some customization.

Mehmet: Do you think like we we have the same Situation that we saw when people started to move away from closed [00:25:00] property operating systems To now we are living in the You know, in the world of the AIs and the big data, do you think it's the same thing?

Ashis: Yeah, I, I completely agree on that. You really correlate to the, the interesting concept that, that was really successful in a time.

Ashis: Um, and this is a synonym to that. So, uh, definitely, uh, like you're saying this, you, you really touched up into one point that. Um, there's some time organization just need, just need something that they don't, they don't care about here and there, or they don't have any, any, uh, any implementation expectation in several years.

Ashis: Then they might go with a proprietary system. Yes, definitely. But in cases where they would like to have the, the. team to grow or they would like to have these days, you know, like, uh, most of the, um, the organization that they would like to be self sustained themselves. Definitely. They need to depend on venture.

Ashis: They need to bring on board the consultants. They need to bring on the, the it companies to have [00:26:00] them. Uh, but they, they would like to be self sustained, at least in a way that they understand what they are implementing. And at least they would like to have, uh, these platform that they understand they, they can control off, you know, not something that it's a black box, somebody, Hey, I'll give you a server and just deploy for me.

Ashis: And I don't, I don't want to worry about you. Would you take care of the support, take care of everything? I just need to implement. And sometime I ask you to do things and you say, wait for months. Then that is something blocking must organize it. But like you were saying, definitely this is also a game changing thing.

Ashis: And even in this is happening, a lot of cases, like something in the bank, there is a open banking concept. So yes, there are, but there are open banking innovation and there are even in the open source world, even the, you know, uh, even, even chat GPT and AI things come up initially chat GPT was being, it is by the, uh, like inner box, let's say it was so popular.

Ashis: But later [00:27:00] down the line, uh, that Facebook come in and hey guys, I mean, you should not be very much limited to Uh, charge EPG where you have to use the proprietary system and you have to pay the, um, the subscription fee monthly. And Hey, you can, if you want to develop yourself, why not? And Facebook come up and brought up the LLM Lama to model.

Ashis: Um, now you can work on it and you can build, right? I mean, this is where the innovation starts.

Mehmet: You just read my mind, uh, she's actually, because I wanted to tell you, first of all, you know, because here I'm relating to where you started the skills. So here, where, you know, Consultants like yourself, you know, and many others will come into the place to implement and, you know, support this open source.

Mehmet: And, you know, because I was about to ask you about the AI and you mentioned, you know, the llamas and, you know, and it looks like even these large language models are going the same, same way. And, you know, Probably by the time now we're airing this, uh, [00:28:00] it's like early May. And people remember two months ago when there was a case, uh, Elon Musk, uh, versus OpenAI, and he was asking them to open source actually the LLM, which Uh, I think he did for grok.

Ashis: Yeah, exactly. Yeah.

Mehmet: And actually he was funny enough. He was telling them, you should change your company name from open AI to close. So, uh, so I think, I think, yeah, because, you know, the, the thing that the misconception of people about open source, I think, actually, it's like, this is the biggest thing that they get wrong is, you know, It's free.

Mehmet: I mean, free in a sense that, okay, I just get it. It's not free actually. You, you, you need to support, but the benefit of keeping it open. And again, I'm giving the example of operating systems because if they didn't have it [00:29:00] as an open source, probably. You know, even the platforms we are using today might not existed, right?

Mehmet: Because, you know, these libraries started to be shared everywhere. Everyone started to build on top of them, something else. So this is for innovation. It's absolutely, you know, something in my opinion, it should be open. Knowledge should be open to your point. And the second thing, And here what I will ask you the question.

Mehmet: Now, large language models, everyone, you know, these are like the mystery. They are black boxes by themselves as well. Now, are you seeing organizations? And I know this is something not easy. Now, people don't get me wrong. When I say build an LLM, it's not a small task. LLM's training requires, like, huge computers.

Mehmet: Yeah, but again, we started to see and, you know, now with the NVIDIAs of the world and what Apple and Google trying to do is, like, even to bring it to the [00:30:00] smartphone, you know, these and they are calling them micro language models. Yeah. So tell me ish, you know, what's happening there? Are we going to see each organization that are developing their own language model?

Mehmet: I would not call it large model . I'll keep it open size. So tell me more. What, what are hap what is happening on, on, on that side?

Ashis: Yeah, this is, I mean, in technology world, you know, uh, this is kind of a, a situation that if, if you start delving down into something like, for example, instance l and m, we were talking about.

Ashis: So there are hundreds of NLM models right now in the market that some of them are open, some of them are proprietary, some of them are, uh, more mixed up, mixed back of them. But in, in, uh, even in the other world, for example, in the application world, it was the same thing as today. So if you have, let's say if it's a fintech application, if you want to have a wallet, uh, develop, or if it's a bank, if it's a telecommunication, let's say they have [00:31:00] the, some of the, um, uh, phone load, phone, phone, uh, uh, roaming, roaming systems and, and so on.

Ashis: There are some of the systems that in the application world as today, it was, it was also the case that if you go in the market and in the Google and try to find out an application, You'll find out something, uh, most in some of the interesting projects that you find out hundreds of similar kind of projects.

Ashis: A lot of similar kind of the tools, the only difference is something a little bit, it may be some, some of, some of the platform has some additional features, some of, some of the platform has some more features, less features, and then so on that you see the market that there being a, being a, uh, like the IT consulting, uh, partner to the organization that we are the one there to select the best out of those tools.

Ashis: Okay, which does feed the right context? So which does feed the right, uh, right of those tech tools and technology organization have or which, [00:32:00] which helps to solve their problems. So basically we are the one therefore, um, and definitely this is one of the challenging in the organization right now. And definitely this is good that every organization or every individuals or research organization and their They're contributing themselves towards this, these products and solution in the market.

Ashis: And at least, at least one another, they're bringing something new. And, and, and where we, we come in and we will, uh, find out, okay, this is the best suited for you. Okay. You go with this, with this model, you go with this model. And it is same thing happening in the data engineering world. You know, like we, we do the data consulting as well.

Ashis: We do have the big data platform and even the open source that we were talking about earlier. So it's not about something that we just plug and play a few open source tools in the market. So like I said, there are Thousands of tools having the same features and need to build a big data platform. You need to have maybe at least 10, 20 different tools together to merge them together to, [00:33:00] to make them work.

Ashis: So we are there to, uh, to select the best tools out of there in the market and then connect them together to form a platform. We call it a big data platform. Okay. So where you can do. Uh, the data engineering, data analytics and data science, AI and machine learning, like even analog models, then you can, and now the other, other journey comes in now you have to implement analog models.

Ashis: Okay. So now in the market, we go there and try to find out what we don't want to reinvent the wheel, something that that's been already built. Definitely. We, we leverage what communities have contributed so far. We just pick up the best out of the best. Okay. Or maybe. Maybe sometimes they don't have everything.

Ashis: I mean, sometimes you have to pick up the one that is more customizable, that is more flexible to implement new feature. And maybe you have, you can, in LLM model, in LLM model context, that you can opt like that, you can train with your data sets, maybe it's a lighter version that you can, that you can, uh, train with the low [00:34:00] resources.

Ashis: Right. Or maybe sometime, uh, you, uh, you definitely have to train with a higher resource as long as your business permits sometime, depending on the business, but definitely we have to select the best models and connect the dots together to form a platform that we call a, a solution for a business that has the best models.

Ashis: That's the how to solve that problem. No,

Mehmet: absolutely. And just one also point. So I don't get misunderstood. So of course, especially because we target entrepreneurs and founders also. So don't get me wrong when I say about the open source, because still there are many ways you can build some commercial applications on top of, of open source.

Mehmet: And You know, just to give people an example, again, from not this time, not from the OS, again, I gave it from the web. So I just fact checked my numbers and I was right. So today, as per, you know, [00:35:00] the time of recording of this episode, 43 percent of the websites globally runs on a platform. Everyone knows called WordPress though.

Mehmet: Both WordPress, they make money, of course, I mean, Automattic, which is the main company. And, you know, they created an ecosystem of plugins, templates, you know, a lot of things that complement this platform to have something, you know, commercial. So when we say open source, and I'm talking here from a pure, You know a little bit entrepreneurship or entrepreneurial perspective startup perspective.

Mehmet: Don't get me wrong I'm not saying when it's open source, it means free. It doesn't mean free. So of course Contribution is yeah, it's a kind of a donation kind of thing if you want to think about it this way But you can still commercialize it and you know, and actually people will pay for it and just i'm giving example from You know, we gave the examples from operating systems.

Mehmet: It came to my mind, [00:36:00] of course, the web. And actually, you know, guess what? And like the programming language, they are by themselves also like open source. And, you know, the libraries are open source. Guess what? I find another interesting statistics just now is close to 79 percent of the websites, they are on PHP and MySQL.

Mehmet: And. MySQL is, which is data actually, it's, it's a, it's a free, uh, database, like it's an open source database, right? So, so this is why to, to get that just so no one misunderstand us. At least like when we say open source, it's actually democratizing. The access to the technology, because you mentioned this, being able to integrate it with other solutions out there.

Mehmet: Let's say you want to build an API on top of that solution. So you want to feed from, to your bot, you gave a lot of examples of the telcos and the, uh, and the banks. So I want to connect my. You know, I don't know, customer experience [00:37:00] application, which collects data. I want to connect it with my CRM system somehow.

Mehmet: So I need this integration. So that become much easy, unify the data in one single platform. And I think this is what, you know, Ashish, you, you definitely explained to us very, very well. Now I want to, you know, ask you, you know, um, little bit about, uh, Collaboration with, uh, with the community. Like, uh, it's something I know you also, uh, believe in it.

Mehmet: So how, how do you foster, you know, collaboration among startups, enterprises and the community to drive innovation? Tell me like a little bit, maybe a story. Uh, and then, you know, we will, uh, we'll be close to, to, to, to finishing.

Ashis: Yeah. Uh, definitely. So the collaboration, it's something definitely, it's a, um, it's a, it's an ecosystem.

Ashis: Without an ecosystem like it, it never runs, right? So [00:38:00] definitely to runs a ecosystem, uh, there should be something, uh, um, a factor that should be propelling, uh, this. Uh, this initiatives, for instance, there, there is an, um, like, let's say, let's say this, this, for instance, it's alarm alarm model, right? Llama model.

Ashis: So Facebook initiated it and they are, they're contributing from the backend with their sound engineers, or maybe there could be some of the other solutions in the market that you see even individual is, is creating an open source project and like living, living the market, but now, now it's comes to the, uh, contribution from the communities.

Ashis: Now, communities should be driven by a few things. One, usually sometimes the organization themselves they promote or they support the solution because they feel, they feel the value in it. Something big organization that we see. But in general, if we see most of the cases that it's being driven by the community itself when they see the value in the product.

Ashis: Let's say if they see [00:39:00] a open source systems that the open source systems, uh, does have a feature, which is not available in the other proprietary systems, or it's too expensive to find in the other proprietary systems. Maybe sometime like you, even you tossed off upon the founders earlier. So let's say, Not all the founders have initial that much of budget to pay to the, uh, to the SAS softwares or maybe to pay to the, uh, systems because sometimes the SAS softwares like ranging from few bucks to thousands of dollars, you know, like they will not be able to, will not be able to, uh, afford that even the big organization like telecom is in a bank that they don't have the, that much of budget to take into the risk.

Ashis: To take to the rest to go to the platform, right? Sometime so this kind of scenario happens. So that's why they would like to do some kind of POCs or proof of concept, or they would like to go in productionization, they would like to customize it and then to go to the production or they would like to go to the directly to the production with the With [00:40:00] open source features available.

Ashis: So it's about you know, uh, it's definitely uh something Uh, a contribution. It's an also way of giving some of the people that some of the some of the people are organized and they just use it and forgot it. But some of the organization they use it, leverage it, improving internally, and they would like to contribute back to the community.

Ashis: And some of the cases, like you're saying propellers, like they are designed to organize and sometimes they contribute. They see interesting feature in the market. They would support their community by contributing some donations or maybe some events happening, some like making some hackathons and so on.

Ashis: But definitely it's also about the motivation to the end user itself as well. And also motivation to the community contributors, whereas contributors are also motivated enough to contribute to the community. Uh, to showcase their skills, to showcase their, because this is a platform to showcase their skills.

Ashis: You know, like, interestingly, a lot of people have got, they got on their job or got on their [00:41:00] interesting idea while they're contributing to the, to the community projects. But interestingly, while saying that, like when you were talking, touching on the earlier topic, the, when any customers or any founders, let's, let's say if any founders or any, Uh, startup founders or any business that they are trying to, uh, build the, uh, big data platform.

Ashis: It's, it's important to, um, it's important to, uh, know that, uh, the, these platforms, there are some of the platforms in the market that they help them to grow, that, that they help them to innovate, and that they help them to validate their, ideas quickly rather investing or waiting something into proprietary, proprietary loop, you know, that is what I'm trying to say.

Mehmet: Yeah, absolutely. And you know, I, of course, this is, it's like kind of giving back also as well, kind of collaborating. And this is what I was just mentioning, you know, about democratizing, you know, technology itself, the reach to technology [00:42:00] by collaboration. I love this. Absolutely. And she's finally where people can find more about you.

Mehmet: Um,

Ashis: yeah, I'm available, uh, in LinkedIn most of the time. Um, I see his first name dash in, in LinkedIn slash or so ideally, um, um, like definitely for more about Delightica definitely can find more in www. delightica. com. The spelling should be on the screen as well to Delightica, Delightica. com. They can find me website and they can find me on LinkedIn and definitely we'd love to.

Ashis: Uh, talk more about how we can, uh, like, help each other, you know, to help to grow. Ultimately, it's an ecosystem. Here as well, it's an ecosystem. Think about ecosystem moment, like, when I'm saying something to call something to your viewers here, um, like, it's about if they come to me and share their problem, And you are helping me, helping me, or we are helping us [00:43:00] to connect together.

Ashis: It's an ecosystem. See, we're, we're solving each other's problem here. We

Mehmet: are 100%, 100%. This is why this podcast exists, actually. It's, it's to, to be able to connect. And if you see our motto is insights, ideas, and innovation. So this is what I aim is to inspire people, giving them insights. Giving them sometimes ideas, connect them with someone who might together work on the next big thing.

Mehmet: So this is the whole goal. The goal is making the CTO show a community, a, I would say an ecosystem by itself that can connect on a global level. And thank you Ashish really for being with me today. By the way, the, the links you just mentioned, they will be in the show notes. So they don't need to search much.

Mehmet: I will make their life easy so you can find the. Uh, the links in the show notes. And again, I should read. I enjoyed the conversation. Like these topics are, uh, you know, beneficial for, for [00:44:00] everyone, for. Business for for founders for everyone because it's the topic of the Uh of of today, you know, and I think it's gonna stay with us for some time now data AI and big data especially as well So these topics are very important and it's good that you shed some light today with us here today And this is finally for the audience.

Mehmet: If you just discovered this podcast, by luck, thank you for passing by. I hope you enjoyed it. If you did so, please subscribe. We are available on all podcasting platforms, and don't forget to share with your friends and colleagues. And if you are one of the loyal followers who keep coming and keep sending me their notes, so much.

Mehmet: very much for doing so. Keep sending them. I read them all. And please, also, if you want to suggest anything, don't hesitate. You know where to find me. I'm more active on LinkedIn, so reach out to me there. And finally, I repeat this every time. And I was telling Ashis now, this is an open platform. If you have an idea, you want to [00:45:00] share it with the rest of the world.

Mehmet: If you are up to something interesting. If you want to contribute by sharing your knowledge, reach It doesn't have to be purely technology anything that can help startup founders and people in the tech space To do something better than they are doing today. Please don't hesitate reach out to me I would be more than happy to speak to you and see if we can have an episode together Thank you very much for tuning in.

Mehmet: We'll be again in a new episode very soon. Thank you. Bye. Bye

Ashis: Thank you