Pod Insights The Brand Called You Scaling Startups & Tackling Tech Debt: Insights from Thanos Diacakis, Software Delivery Coach & Fractional CTO About this episode Ever wonder how technology leaders adapt between nimble startups and tech giants, or what it truly takes to ship software that matters? In this insightful TBCY episode (S7 E062), Ashutosh Garg sits down with Thanos Diacakis—software delivery coach, fractional CTO, patent-holder, and independent consultant—to unpack 25+ years of lessons from the front lines of engineering and startup life. Below you’ll find key topics explored in the episode, organized by timestamp, complete with guiding questions to help you navigate the most relevant insights for your journey.
Transcript 0:08 Speaker 00 Welcome to another episode of the brand called you a podcast and podcast show that brings you leadership lessons, knowledge, experience and wisdom from thousands of successful individuals from around the world. I am your host, Ashutosh Garg. And today I'm delighted to welcome an accomplished professional from the world of technology from talking to us from California, but today in Greece, Mr. Thanos Diacakis. Thanos, welcome to the show.
0:36 Speaker 01 Thank you very much for having me. I'm super excited for our conversation today.
0:39 Speaker 00 Thank you. Thanos is a software delivery coach and a fractional Chief Technology Officer. He's also an independent consultant. Thanos also holds seven software patents and degrees.
0:51 Speaker 00 So Thanos, looking forward to a fascinating conversation with you. And my first question is, what drew you into software engineering originally? And what keeps you passionate about it after 25 years?
1:04 Speaker 01 Yeah, it's almost too long for what really brought me in. But I remember just as a kid being super into anything that has buttons and knobs and switches on it. And I was just super interested in learning how these systems worked. It was like figuring out a little puzzle of how do you work with this thing and make it do exactly what you want. And it really got a lot of satisfaction out of building something out of nothing. And I think that continues to present day where when I go work as an employee or now as an owner of my own business, I really enjoy working with clients and actually coding stuff. I see those little tiny problems and you have to solve a lot of these little problems together. And this is tiny puzzles. I get immense satisfaction out of solving them
1:47 Speaker 01 and using your brain. And you start from like me applying it to a technical problem that is like debugging something, but then a little later you're debugging a team or a process. And it becomes a lot more fun is looking at these bigger systems and how they all come together.
2:01 Speaker 00 Fascinating. And you worked across startups and technology giants. What have been some of the biggest mind shifts you had to make between those worlds?
2:12 Speaker 01 Yeah, these worlds are like super, super different. In a startup, you're little and scrappy and you kind of work in a bit of everything. So my first startup was out of college and I was VP of Engineering and very little I knew about managing people. But you get thrown into the fire and you have to learn really quickly as how to do a lot of different things. And it's not just how to build software, but it's about how to do all the other little bits and pieces around them. And then you fast forward, you know, a decade or two and you go work for like a big giant like Uber. And you no longer have to do everything yourself. And before you were working as a piece of software and you needed something designed, you would design it yourself. Or now you have a designer to go to. So there's a lot of differences in the amount of resourcing that you have and how you work with people.
2:56 Speaker 01 And the way that each of these gets scaled is super different. Like in a startup, you get scaled by not having any process and you're just running as fast as you can towards the target. In a big company, you get scaled through collaboration. You get to work with others. You get to like leverage a lot more people than you would at a startup. Each with their individual expertise. And from there, the scale comes from that collaboration. So there's a lot more different activities where you know, you have to get consensus building and you have to get everyone to buy into what you're doing. So it's a lot of fun of seeing how it works and adjusting your modes depending on if you're here or if you're there or how are you going to play any particular aspect of things.
3:35 Speaker 00 Amazing. So let's talk about startups and scale up. You know, you do a lot of work with startups and scaling up. So we'll focus the conversation today on this topic. And my first question is, what are some of the most common engineering bottlenecks you see in early-stage startups?
3:54 Speaker 01 Yeah, for me, tech debt is the one that keeps coming up time after time. And for those who are not familiar, the idea of tech debt is that when you're building something new instead of building it the right way, you might take a little shortcut and incur a little bit of tech debt. And then later you go and pay back that tech debt and fix it. And you can have this cycle back and forth. But frequently what happens, this is unlike like a line of credit where you borrow, pay back, borrow, pay back. But it ends up being like sovereign debt where a country borrows and borrows more and borrows more. And it ends up with the IMF having to step in and the country going bankrupt. So a very similar thing happens with the startups where they will borrow, but will not pay back from this tech debt bank. And you know, it's not necessarily a bad thing to use some tech tip, especially if you're
4:39 Speaker 01 really early on startup. If you don't build quickly enough, you may not be there to find out what happens after that. So it's okay sometimes to take shortcuts. But you also need to come back and take care of these issues. And you need to have like enough experience of what are the right kind of shortcuts or what are company killing shortcuts, right? So if you, there's some mistakes that you cannot fix later. Those are the mistakes that you can say, you know what, I can make this design this way or that way. That will go quicker, but it's in this box over here and I can go and take care of it later because it's compartmentalized, suppose. So that other thing that will affect everything and it'll be really, really hard for me to fix later. So striking that balance between how fast you go and how right do you build things is one of the things that startups often struggle to get right.
5:23 Speaker 00 Very interesting. And you know, when you work with founders, one of the things I'm, I've been a founder of two startups myself. How can founders recognize when their engineering processes are holding them back?
5:38 Speaker 01 Yeah, so that's actually really quite tough. And I would expand this though. I would say it's not just processes. You have three things in there. You have process, you have culture and you have technology. And each of these things are a little bit unique and it's, all of these three could be holding you back. The way I work with founders in figuring this out is, is figuring out where your bottlenecks are. And this has involved, involves them taking a step further back and seeing what are you trying to achieve? And a lot of the times you'll see founders that will focus on achieving what I call input metrics, like did we do this amount of work? And oftentimes, you know, that's nice, but it's not the thing that matters the most. What really matters is what did you ship? Did you add value to your clients? And that's what I call an output metric. So
6:20 Speaker 01 I really would prefer to focus on your output rather than what your, your input is in this case. So we define what the output is, which is getting things out the door. That usually involves slowing down. It's like not starting too many things, but finishing enough things. And then working backwards from there and figuring out what the bottlenecks are to get the outcomes that they want and fixing the bottlenecks and removing the bottlenecks. And then you do this over and over again. And sometimes the bottlenecks are process, like could we do this thing different? We all agree, we just do something different. Sometimes it's culture, like a new employee comes in, they see how everyone else does things and they do the same thing. And that could be a good thing, could be a bad thing. And then the leaders need to communicate and figure out how to fix that problem. And sometimes, going back to tech debt, we may have some bad choices that we made, or maybe we took some shortcuts, we need to fix those.
7:05 Speaker 01 And, and, and go from there. But the whole process of, you asked the question of how do you recognize is, recognize what you're trying to do first.
7:11 Speaker 00 Uh huh.
7:12 Speaker 01 Be very clear that the goal is shipping value to end users by way of software usually. And then work backwards from there to figure out what the, what the bottlenecks are.
7:20 Speaker 00 Hmm, well said. And what is your framework to align technology strategy with business outcomes?
7:28 Speaker 01 Yeah, that usually involves looking at the interface between business and engineering. And I think a lot of times that interface is, is very misunderstood. You think like if you add more engineers, you get more speed. Um, and that's not necessarily always the case. Uh, so I start with a set of mental models. And I have developed a few, uh, and this involves thinking about engineering differently. Thinking about the interface of engineering differently. I'll give you a couple of examples. Uh, one of them is tech debt. And, uh, we talked a little bit about tech debt earlier. Um, having the business side understand that more than one thinks. They don't, don't just build features, but they also have to fix bugs.
8:12 Speaker 01 They also have to go back and improve and sharpen their tools. They also have to adjust with risk items. That then, you know, makes some light bulbs go off and say, oh, so if we ask for like 90% of your time to be going into features, that's not going to be a good outcome for the long-term velocity because other things are going to acquire sort of rust into them.
8:32 Speaker 01 So, once we have these understandings of these, these mental models, I think now we can start looking at what are the actual solutions to these things? And it's, I think the mental models are important because we can't hand you a checklist, right? If I hand you a checklist, then you're going to need to adjust the checklist at some point in time because it's not going to fit exactly. And that will usually not end up well because a checklist is a very narrow prescription. And I want to give people the tools where they can have a bit of a broader understanding of what we're achieving. And what are some good principles and some bad principles? And then you can adjust your checklist as you go. So, once we have the mental models kind of like figured out and applied to our situation. I kind of follow like a four step process, um, which is iteration, quality, complexity and planning.
9:14 Speaker 01 And I'll tell you a little bit about each of these, uh, real quick. Um, and, but before I do that, I want to say that these are, I've made these linear because it's easy to get your head around and understand these. They might not always be linear, right? You might not always start from step one, you might jump a little in step two and then jump back and forth depending on your specific situation. However, if you're starting from step four and working your way fully backwards, you're probably doing something a little wrong.
9:37 Speaker 00 Yeah.
9:38 Speaker 01 Um, so it's, you know, it's, it's linear as a guideline. Um, and the first step in this process is iteration. This is your ability to go from an idea to software running in production in users' hands in a fairly short period of time. This is in days, not in weeks or not in months, like it used to be 20 years ago. Um, once you have that one down, you go on to quality and production. This is understanding how your system works in production. Do you understand what it's really doing? Does it do what you think it's doing? Because sometimes you think you know what it's doing, but it's not doing exactly what you're doing. This is observability, all these kinds of tools, um, that enable you to see what's actually happening. And it ties in together with iterations because at some point you cannot iterate fast enough unless you can see that you're not breaking what was already there.
10:19 Speaker 01 So now, once teams have iterations down and they have some pretty good level of quality, what tends to happen next is complexity sort of starts growing up and goes higher and higher. So then we have to get that under control. And that's not always architectural complexity, which is the first thing we tend to think about. But this could be procedural complexity or even cultural complexity in some ways.
10:38 Speaker 00 Right.
10:38 Speaker 01 So, uh, this complexity builds because, you know, you keep adding stuff and adding stuff. And then you have to figure out what, what are the relationships between these things to each other? Are there any things that we can remove now? Are there things that features we built, but no one is using? But every time we have to make a change, we can break that feature. But if no one's using it, we just remove the feature and then go faster in other areas.
10:58 Speaker 01 So, so working through this complexity, for me, complexity is like a very serious enemy because we are limited. Our human beings, as smart as we seem, we have lizard brains. And at some point, there reaches your capacity of processing and it becomes really, really hard to scale things. So we have to keep complexity down. And finally, when we nail complexity, now we got iterations, we have quality, we have complexity nailed down. I think then we reach the point where we can start thinking about planning and predictability. And I've put this intentionally forth on my list because a common mistake is people try to reach for planning and predictability in the beginning. Things take too long and then they want to, you know, kind of put some specs down and try and figure out how to make this much more predictable and we can break every little step in the plan and we're going to execute and make sure next time it goes exactly to plan. And that really never works out well.
11:43 Speaker 01 It's better to have a way to be adaptable and flexible and be able to zig and zag every time something happens than trying to be rigid. But it's also not really good to try and plan and be predictable if you're not really good at executing. If you can't iterate, if you can't get your quality quite right, then you're not going to be able to, you know, figure out how really complex project is going to take. I joke and I say it's asking me how long is it, you know, when am I going to get to Mars? When I have a horse and a buggy and it breaks down every half mile, right? I need to first be able to, you know, you know, take the horse and buggy, make a trip to the next town, then build the car, then build the plane, then build a spaceship until I can tell you how long it's going to take to get to Mars.
12:20 Speaker 00 Yeah.
12:20 Speaker 01 So we build upon the capabilities and get the planning and predictability. So I like that force that framework because it gives people that are lost or have way too many problems a way to get started and start picking through these problems and get some progress and see some results. Because that's encouraging. You get, you know, you deal with something, you get some results. Now you're happy to work on phase two and phase three without sort of losing any hope there.
12:42 Speaker 00 Great response. Thank you. What qualities do you look for when helping a startup scale its engineering team?
12:51 Speaker 01 Yeah, that's a super interesting one and it's like really changing right now with AI. So my answer is going to have a little bit of an AI component. One of the things that I look for to start with is I like people that are good engineers. And for me, engineers is not knowing a particular language or framework or system. And those are nice to know, but I want people that know the general concepts of engineering. How did they make trade offs? How do they solve different problems? How do they learn new things? What are their approaches to problem solving? Usually when I see a resume and it says like I'm a Python engineer. A Rust engineer or I'm a JavaScript engineer, that's almost immediately goes into my reject pile. Because I don't want people to know a specific language. I want people to know how to learn a new language, how to understand new things.
13:35 Speaker 01 So that's one of the first things that I look for or advise my clients of how to how to filter as they as they look for hiring. Another thing that I look for is I want engineers to think of technology as a tool to meet user needs. This is not like, I mean, I love technology. I'm like a nerd's nerd. I love to get in the esoterics of memory management and how you do very specific things. That's fun for me, but your company, your project, your team is not going to win because you picked like a really cool technology to go solve the problem.
14:06 Speaker 01 You will win because you are able to iterate quickly and solve a user problem, receive feedback, learn quickly, and ship something that meets a customer need. So I try to suss out how people think and how people are interested and motivated in solving problems for their client bases, or are they super interested in just using a particular technology, or they're going to be really rigid in using a particular way of solving a problem. That usually ends up in just having debates that never end and doesn't end up really good. And I also look for people that just like getting things done. I don't want someone that just wants to get tickets and they can ship the tickets and get new tickets. I want some people that like to collaborate, they enjoy the process of it, they want to discover things. And the last one that I'll say with with the related to AI is we have this new tool now.
14:50 Speaker 01 And I want to find people that are not afraid of the tool. And I'm not going to use AI. People are going to be better engineers. Or they're going to be like, I only use AI. I don't even look at the code. I just use AI. I want people that understand its limitations. They understand how to use it. They can be good wranglers of the tool. And they know how to use it and they're willing to do so because I think the people that are the best learners are going to be the most successful. I used to play this computer game that said, it was a flight simulator when I was a kid. It said, one of the advice that was written in there was know and use all the capabilities of your airplane. If you don't, then sooner or later, someone who does will kick your ear.
15:29 Speaker 01 So for me, that's the same with technology, right? If you don't learn and use everything that is at your disposal, someone else will do so and they'll take your lunch.
15:36 Speaker 00 Very well said. And based on the work you do with so many startups, my question is, when should a startup invest in infrastructure and performance optimization?
15:48 Speaker 01 I'll have different answers for each of these. Infrastructure for me is pretty straightforward because it is a multiplying factor. If you have good infrastructure, then your engineers can use that infrastructure and do common things that most engineers do, and they will save a lot of time. You will get a lot of mileage out of infrastructure. A lot of the clients that I end up working with, we fix a lot of infrastructural problems, and you can see like 2x, 3x, 4x gains in speed because it just unlocks the engineers from doing this. I was working with one client that didn't have a type checking system in their code base. So every time you try to write something, you just make silly mistakes because some type didn't match and you didn't catch it. So we worked diligently to fix this problem, and boy, now they're like four, five times faster as they develop any kind of code.
16:33 Speaker 01 So I would say early on with your infrastructure. Again, this will depend on your resources, but that is good return on investment usually. In terms of performance, I would tend to say performance comes a little bit later, right? You kind of want product market fit with what you're doing. So if your product's a little slow or a little buggy, usually you can tolerate a little bit of that early on, and you can deal with fixing those things a little bit later. Go back to Twitter days, right? Twitter was down all the time, but that did not prevent it from becoming a great service because they can find a product market fit and then fix these things later. So, you know, don't do things that are like company breaking in the sense that they will not, you will not be able to just scale to any point of user basis, but premature optimization
17:17 Speaker 01 is a very common pattern where you can make this the fastest thing, but that is not necessarily the problem that you have.
17:23 Speaker 00 And Thanos, what is your approach to build versus buy when it's time, when time to market is very critical?
17:31 Speaker 01 Buy.
17:32 Speaker 00 Okay.
17:34 Speaker 01 I'm half joking in being so absolute about it, but I think it's very tempting for engineers to go and build stuff because it's cool to build stuff.
17:41 Speaker 00 Yeah.
17:42 Speaker 01 And 90 something percent of the time, that's a mistake. If something is a commodity, go buy it. It's a little bit tricky, so there's a little bit of nuance here because sometimes when you go buy something, you will get the last 10 and the last 20% problems. And the last 20% problem is you'll get from what you bought 80% of the functionality and now it works just great. And the next ten percent going from eighty to ninety percent completion is gonna be tough because it's gonna be off the shelf and it's not gonna do what you want. So it's gonna be a real challenge to integrate. And you might spin a lot of your wheels and try and get that integration done. And the last ten percent may be impossible. So that thing just does not do what you want and you would have been better off to buy.
18:24 Speaker 01 So that is a little bit of a tricky choice sometimes and I think that's why we struggle it. So I tried to put it in this framework so we can understand it. Um, and, and act on it sort of explicitly rather than implicitly. Is you will want to do your diligence beforehand to figure out the ten and twenty percent problems. Because if you do that, then your hands are unlocked. Um, the general inclination and advice that I have for things is just go try something out, right? Because by trying it out, you're gonna learn really quickly. So if you're able to take a third-party product and test it for like a few weeks or a few months or draw a contract that says we're gonna test it out. You can see if you have that the last ten or twenty percent problems. And if you don't, great, go buy it and knock yourself out.
19:05 Speaker 01 But if you quickly hit these problems, then you have some clues and some areas of research where you say, okay, are we okay with that? Can we live without the last ten percent because buying it's gonna be so much cheaper? Or do we wanna go build it, but now we have to, it's like a puppy, right? You have to feed it and you have to take it out and you have to clean it, take it to the vet. So it, it, there's a lot more maintenance that, that goes with that. So that's kind of like the framework that I use to evaluate it typically.
19:29 Speaker 00 Well said. Uh, time for two more questions. What role should engineering leadership play in shaping company culture at startups?
19:40 Speaker 01 I would say that is possibly their only role or one of the most important roles that engineering leadership has. You, you probably have heard of the old adage that, um, culture eats strategy for breakfast, right? And, and the idea is that if you want to get something done, you just post a new strategy. That's just not gonna happen if your culture is the exact opposite of that thing.
20:00 Speaker 00 Correct.
20:01 Speaker 01 And, and I had this with a recent, um, project that I was working on where the culture was just made a Leadership wanted to go a different way and they just said this is what we're gonna do. And it took over a year and massive changes to start changing the culture before we could get to that point in time. So I think when you get to the top, when you have a VP of Engineering or a super senior engineering role, you lose a lot of levers. Like you don't write code every day, you're not in every meeting of the company. So the biggest lever that you have is communication. You can communicate a strategy and you can communicate it over and over and over again. And you can by example demonstrate some cultural aspects. But the idea here is that you have to build this culture that is gonna be able to execute your strategy.
20:46 Speaker 01 Because otherwise, I think you're just gonna fail. You're just, you know, you're gonna be hitting a wall all the time. When the culture is opposing you and nothing, nothing is really happening. So as, as, as, when I've had leadership positions, I think for me, that was a super important thing of communicating something over and over and over again. And people are busy, people have their jobs to do. So just when a leader comes up and says something once, not necessarily everyone heard it, not everyone paid attention to it. It's yes, one, one more of these things that Thano said. So it's like communicating consistently and over time and bringing things back and repeating them is, is key to this thing. I think there's a saying that like most things, you have to say them seven times until they stick.
21:25 Speaker 00 Yeah.
21:26 Speaker 01 So, so in, in that kind of environment, maybe you say it like 70 times or 10 times or something like that. So.
21:31 Speaker 00 Amazing.
21:32 Speaker 01 Culture is super important.
21:33 Speaker 00 Yeah, great response. And my last question, you know, we've spoken about technology, we've spoken about startups, we're talking about, spoken a little bit about technical debt. What advice would you have for startup founders trying to scale fast without losing control of their code base or their culture?
21:53 Speaker 01 So, so let's pick those apart, right? Because I think code base is a super interesting one, especially now with the age of AI.
21:59 Speaker 00 Yeah.
21:59 Speaker 01 Because if you don't have the right processes in place, the code generated by AI will go off the rails really, really quickly. So this was always important, but now with AI, it's even more important that you need to have the processes in place to be able to maintain quality. This, I mentioned in the beginning, iterations, you need to have the processes to have iterations in there. That involves controlling your code base, making sure that certain practices are followed from linters to compilers to all that sort of stuff that you have in there. You know, how are you doing code reviews if you're doing code reviews? So again, if we thought the bottleneck was like typing in the code, well, The bottleneck, the bottleneck is do you have all those systems in place to guarantee
22:41 Speaker 01 the sort of qualities that you want from the system to exhibit the right characteristics in there. So that is a super important aspect and leaders need to super stay on top of that. I have a feeling that as there's more and more AI-based startups that they just let the AI run loose, that I will have a lot more business going in these startups in the future and cleaning up some of that stuff. I mean, it's a really nice tool, but like you have to show it like like nature, you need to treat it with respect and understand its limitations and deal with it.
23:16 Speaker 00 Correct.
23:17 Speaker 01 So I think you have to stay on top of that. In terms of culture, I think that goes back to mostly to my my previous answer. That is the primarily rule role of a leader. You, your culture is shaped by what you do, how you set an example, and it's also shaped by things you tolerate. If you say, yeah, we tolerate that person's not a very nice person, but we tolerate them because they get good work done. That is now your culture is allowing not very nice people to work in your company, and that's going to have adverse effects. So like that that edge is what defines your culture, and I think leaders should pay a very close attention to it. Is that that is company defining.
23:55 Speaker 00 Absolutely. And on that note, and you're once again highlighting the importance of culture, which I think drives so many different businesses irrespective of what line of business they may be in. I want to say, Thanos, thank you so much for speaking to me about your own amazing journey. We covered some very interesting ground for startups on technology, on culture, on, you know, for early-stage startups, for founders, and of course, for technical debt. Thank you for speaking to me and good luck to you.
24:26 Speaker 01 I very much enjoyed the conversation. Thank you for having me.
24:29 Speaker 00 Thank you. Thank you for listening to The Brand Called You, video cast and podcast, a platform that brings you knowledge, experience and wisdom of hundreds of successful individuals from around the world. Do visit our website, www.tbcy.in, to watch and listen to the stories of many more individuals. You can also follow us on YouTube, Facebook, Instagram and Twitter. Just search for The Brand Called You.
Previous Tech to Transformation | Jay Sehgal, Exec VP, Sehgal Foundation, USA & Chairperson, S M Sehgal Foundation Next A Life of Grit & Grace | Yalda Aoukar, Co-Founder and Managing Partner, Bracket Capital Page updated November 15, 2026 · transcript and summary are machine-generated