Solving for data intelligence at scale

Madalina Tanasie’s career has been led by her belief that technology should be purposeful. Embracing choices that reflected this conviction, her professional path ultimately led to the C-suite.

David sits down with Madalina to discuss the always-evolving technology landscape and the recent advancements in AI, especially in the context of her work at Collibra.

Join us to discuss:

• The need for engineering leaders who can translate technical issues into business pains.

• The “buy vs. build” mentality — and the benefit of leveraging managed services.

• AI governance and the importance of regulatory compliance and informed business decisions.

David Joy:

What is up, everyone, and welcome back to this new episode of the Big Ideas in App Architecture podcast where I speak to Madalina Tanasie, who is the CTO of Collibra. And I and Madalina get into some really great conversations around her career, her experience at Medidata and what she is doing at Collibra as a CTO, getting into all the amazing things that her team is working on.

We get into some of her philosophies around engineering and how she’s building teams that are solving interesting problems. One of the most exciting things in this episode is her perspective on AI and AI governance that their team are considering and their users need. So, jump into this episode and have a great time listening to Madalina.

All right, Madalina, welcome to the podcast. How are you doing today?

Madalina Tanasie:

I’m good. Thank you for having me.

David Joy:

Let’s start off. For everybody listening, Madalina here, she is amazing. She’s working with Collibra. She is handling some really critical things for the company. So, before I butcher what you do, tell us a little bit about yourself and what you do at Collibra.

Madalina Tanasie:

So, a little bit about myself, I think of myself as a mission-driven technologist. About my role at Collibra, I’m the chief technology officer at Collibra. I’ve been in this position for three years. Again, very important for me is that technology for a purpose. And from the very beginning, even before deciding to do computer science, I was fascinated about the potential of technology to solve some of the biggest problems.

And more by accident than having a very intentional path in terms of my career, I made choices that were aligned with that sentiment, that technology is for a purpose. And I found myself a few years ago really looking at Collibra, who is trying to solve some of the big problems that we have right now. We have an explosion of data. The biggest companies out there have an incomprehensible amount of data coming every single day.

And everybody’s trying to make sense of their data, and we are trying to make sense of all the data that assaults us. Can you imagine how that is for the biggest companies? Collibra is the largest SaaS provider for data intelligence solutions. We do have a platform that covers all kinds of data solutions, from cataloging and governing your data to lineage, data quality, privacy, you name it.

And it’s particularly exciting for us right now with AI and how that’s changing everything that is done in this space and how data is used and what is used for, but also really just taking this idea of governing data to governing for AI and AI governance.

David Joy:

Yeah, no, that’s pretty cool because you’ve brought up, everybody knows, and especially in the last decade, we’ve been saying data is the new oil. But recently, in the last few years, we have figured out how to tap into that oil and pipe it out and turn that into something. And it’s exciting. And four months ago, when I was talking to you, AI was at a different journey.

We were at a different point. And in four months, there have been so many things that have happened, that’s like we just have to be-

Madalina Tanasie:

I was actually thinking because when we spoke about this the first time, there was something there, but I feel like it’s almost good that we waited with this by chance because I feel like so much has changed and it’s changing so fast. It’s almost every day it’s something new, and we’ve never been in such an accelerated reality in tech.

David Joy:

Yeah, I agree. This is the fastest I have seen things change. And I keep in touch with most of the things that happen. I have written my own agent that scours and scrapes the internet and summarizes the latest information. And what I’m seeing every day is that agent fails actually. You know why it’s failing is because when it tries to scrape everything and put that into a JSON and tries to load it, it fails to load because there’s a threshold that it-

Madalina Tanasie:

There’s just too much data.

David Joy:

… times out. Yeah, it times out, and it’s insane the way things have changed. But anyways, I really am excited to talk to you about all the amazing things that you have been doing. But you’ve had a fascinating career. I was just going through your LinkedIn, just reminding myself. You’ve worked in Medidata for 15 years, which is something that we should get into a little bit.

But tell me a little bit about your early career. How did a young Madalina decide to like, “Hey, this is what I want to do.” What inspired you to get into tech?

Madalina Tanasie:

Given where we are, a lot of people are curious about how I did it because there is this assumption that you have this grand plan and you are super committed and you execute on that grand plan. And as I was saying earlier, a lot of it fells by accident, but it’s not really that much by accident as it’s having a purpose and having that guide you. And my journey has been very zigzag.

It’s been sometimes really straight and sometimes going sidewise. But I had no plan to go into computer science. I actually was preparing to go to an architecture school, not software architecture, buildings architecture. And I spent a couple of years in high school going tutoring and everything related to learning about buildings and designing buildings.

I was really fascinated by beautiful things. And I realized, as I was going through that journey, that I’m not really good at drawing. I can be decent, I can be adequate, I can never be really, really good. And for me, it was… I have some skills, some undeniable skills. I’ve been pretty good at math, physics, science, always kind of ahead of my peer group.

And there was this expectation that I’m going to do something with these skills, and I wanted to do something with those skills. And at that point, computers and internet and technology was kind of picking up and it was the latest and newest thing, and everyone was saying this is the future. And it’s like, okay, what do you need to get into the computer science course? Like math, physics, I can do this.

So, kind of went there, passed the exam. There’s super high competition in Bucharest at that time for computer science. Everyone wanted in and I just got in. And I didn’t even have a computer before going to college to do computer science. And I loved it. It was the right choice, but it was a very, very last moment choice. Mostly focused on what can I do that I can be good at?

David Joy:

It brings all these skills together. That’s amazing. To know that somebody is a CTO today, “I chose this career and didn’t have a computer to begin with,” was just inspired to bring. It’s a fascinating story. How was the computer scene, the science scene in Bucharest, in Romania when you started off? Was it something that was picking up in the country? How was that for you?

Madalina Tanasie:

It was emerging. I actually went because of those skills that I had recognized. I went to a technical high school. I had some computer lab classes, part of my high school education. So, it’s not that I saw a computer first time in college, I just didn’t have my own until then. By the time I finished college, there were a few companies, startups, that were based in Bucharest and other big cities that were starting to get into a field but was really, really small.

Just a lot of technical STEM educated people getting into it. So, it was very, very interesting.

David Joy:

Got you. That’s amazing. It’s great to know how somebody starts the story, and I love to hear people’s story because it inspires so many people who listen, who are on a different journey and different points in life. So, thank you for sharing that. I was curious now to get into a little bit more about, from there on, you’ve had a significant career.

You spent almost 15 years at Medidata, starting as an engineer and then you took management roles. So, tell us a little bit about those beginnings and what kind of shaped your early part of the career.

Madalina Tanasie:

Yes. So, I had no plan whatsoever to go into management, let’s start there. I didn’t have a great plan. For me, it was, “Hey, I want to pick a company or a role where I can help with something that is important to me.” And Medidata, I don’t know how much of your audience knows about Medidata, but Medidata has the largest platform for clinical trials.

So, a lot of the companies that are inventing new drugs and new medical technology are running the clinical trials on Medidata’s platform. And back then, Medidata was actually really, really small. I was employed 247 or something like that. So, not tiny small, but small comparatively. And they had a single product. It was the beginning of a big great journey.

I liked what they did, the skills required for the job aligned with what I was doing. And I applied and they got me. But of course, as I was saying, for me, it’s always been technology for a purpose. So, I was never really satisfied with just doing what was asked of me. I kind of started working around my scope and expanding it a bit and putting two and two together and coming up with ideas and proposals.

And my then boss noticed that I’m interested in that space and gave me a very unusual goal. It’s like, “Your year-end review is going to depend on you achieving this one thing. I need you to be known by one executive that you do not know. When I go there and we do calibration or whatever, and I say your name, one of the people on the executive team should know who you are.”

And it’s like, “This is ridiculous. No, who am I? Software engineer X on this corner of the floor speaking with executives?” But I’ve always been good with challenges, and I don’t even remember who I approached. But by the end of the year, they knew my name somehow. And it was just such an interesting revelation for me. I feel like, ultimately, it’s been the single biggest differentiator for me and I think could be for anyone that is doing technology for a purpose.

Realizing that being able to explain deeply technical topics to business folks is what’s going to tremendously increase your ability to make an impact. Because for many people, the ones that decide, the ones that deal with the money, the ones that make the business decisions, what gets done, what doesn’t get done, technology is this black box. And they understand about something but not enough to feel truly comfortable with making decisions.

And if you’re able to really just take something that is abstract and explain it to them in a way that they get it, your ability to impact is tremendous. And that’s what I realized with that crazy goal. And I feel like it’s been something that has continuously pushed me forward because when I found something that I cared about, I found a way to explain what I had in mind to the person that made the decision.

David Joy:

Yeah, no, that’s amazing because when you were saying this, I have heard this from another person, exactly same, quote by quote, this idea that this is actually an art and not many people can do it. But there are few who have been gifted with this idea or this ability to look at a complex technical problem and know how that works, and then distill that, simplify that and talk about it to non-technical people who also need to make decisions. So, that’s what you’ve discovered.

Madalina Tanasie:

Yeah, absolutely. And it’s really that idea, and you’re right, it’s a gift. Being a technologist, often you’re with hands and head and everything in technology and you’re not really thinking about people that don’t get it in the same way you do. But for me, it was the reality that I had that gift, but more importantly, that someone noticed and put two and two together and challenged me to actually have that interaction and how do you go as an engineer and speak with an exec that is not on the technology.

You have to figure out how to explain to them in a way that they get it. So, it was a big interesting revelation for me.

David Joy:

Yeah, I have had those moments in my life where I have to do some things like this when I’m mostly on the business side of things, but I have major withdrawal symptoms. I need to code something. So, how I compensate for that is either I will find a project, a POC, work on it within the company, or I’ll file something outside. But because I am also somebody who likes to be in these two worlds, but for some reason, I have major withdrawal symptoms.

I do need to look at code and do some coding and things like that. As you moved up in management, did you continue to also stay on top of technology, worked on tech code level?

Madalina Tanasie:

Most definitely. In my career at Medidata, I was very, very close to the code and action, really all the way to the end of my career there in part because I was the original engineer on a lot of the products that I ended up leading that became foundational for Medidata’s platform. And I was still the one that knew all the skeletons in the closets and could trace back many of the decisions.

I’m a little bit less close to the code at Collibra because a lot of the decision predates me, but also the tech stack is different than what I’m comfortable with as an engineer. That’s not to say that I’m not in Git looking at people’s code more often than they are comfortable, and I’m not in Jira commenting on requirements more often than everyone is comfortable.

So, maybe that’s kind of how I deal with the withdrawal. Sometimes I get bored and I go really, really deep, and everyone is freaked out. It’s like, “What happened?” It’s like, “Nothing, nothing. Just like-”

David Joy:

Why are you looking at Jira code?

Madalina Tanasie:

“Why are you looking at Jira? Why are you looking at Git? What happened? Why are you looking at logs?” So, I do some of that.

David Joy:

Yeah, but you also earned it, right? You’ve also earned it. You’ve been in the space. You’re a veteran, you understand how this works. Now, as a chief technology officer, of course, your roles and the objective and the outcomes that you’re driving is different from what you’re doing as an engineer. So, I wanted to dive into that a bit.

But you brought up a great point about Collibra, the tech stack and the tech stack that you have and were working at Medidata. Tell us a little bit about how these two were different. What are the challenges that you were trying to solve from a technology point of view?

Madalina Tanasie:

The part that is very, very similar between these two companies, and it’s not the same tech stack, let’s start there. The part that was very similar is that when I joined these companies, both of them were in their second decade of existence. And when you are there and you work in the enterprise space, the customer base for these two companies are the largest companies out there, global 500.

In 10 years, you’ve accumulated a lot of technology. There is no such thing as “Hey, this is our tech stack.” You have built all kinds of things. You’ve started when the norm was something and then now the norm is different and you continue to evolve and you continue to build things, and you have cacophony of technologies and patterns and whatnot. Also, both companies had, by the time I got there, a number of acquisitions.

And when you acquire companies, you’re going to have different stack for those companies. So, I think the part that is very interesting for both of these companies is like, “Okay, what do you do with this? How do you evolve something that is very complex by the nature of just how the company got to be successful, but also complex because the customers that you’re dealing with are really complex and have needs that are difficult sometimes to get uniform and say, ‘I’m building this and it’s going to work for all the customers.'”

So, in enterprise software, it’s a little bit different. I think the part that was very interesting about Medidata is right as the cloud was becoming something that everyone was starting to look into and service-oriented architecture was becoming mainstream, the business needs of Medidata truly aligned with those architectural patterns. We wanted to start to move from just electronic data capture and the one product that Medidata had to a platform.

And we recognize that if you want to give something to the providers, the hospital doctors and nurses, to work on multiple clinical trials across pharmas, you really need some of that multi-tenancy. You don’t want them to take the same courses multiple times. You don’t want them to have multiple logins. If they are going to use Medidata software, they should log in once and see all their clinical trials.

And if they are to be trained on Medidata’s EDC module or whatever module, they should take the Training 1. So, there was this need for multi-tenancy and then there was this potential of cloud. So, it was just a fascinating moment where what was becoming like this gold star in terms of architecture really aligned with the business need.

Now, fast forward to where we are now, a lot of these patterns have become mainstream, but I think we’ve lost a little bit of that, oh no, the word is not coming to me. It’s like that lean thinking to recognize that some of these patterns are really not just applicable in every single case. And if you start a project really planning for the most scalable, most architecturally beautiful thing, you’re probably going to overdesign.

We have something super expensive, difficult to operate. And I think at Collibra, we have some of those needs, but because we are actually working across industries, across domains, there is very little of that repeatability that justifies multi-tenancy. A lot of multi-tenancy, it’s around technical components, more than just functional components.

So, technology decisions, in my mind, going back to what I said, I’m pragmatic, mission-driven technologist, are first about what does the customer need and what is good for the business. It’s not about asking can we do this? Yes, if you have a smart team of engineers, you can do anything. I’ve learned that. I worked in a number of programming language. I can read more than I can write.

Smart engineers can implement anything you throw at them. That doesn’t mean that you should do it. The right question is not can we do it, it’s should we do it? Can we build a sustainable business model around it? Can we provide real value, and can we do it in commercially sensible ways? We spend tremendous amount of money to build it and to operate it, probably shouldn’t build it.

David Joy:

Yeah, this is such an extravagant, amazing response. You’ve touched upon so many different ideas and also I can see how you as a CTO are looking at things when you come and build solutions. So, when you were talking about Medidata and the early cloud, I want to dissect this response of yours and go at each of those. I’m curious, what was the cloud you guys first started exploring and how did the solution?

Madalina Tanasie:

Right. We were early adopters of AWS. And for the longest time, we were really just in lockstep with them. They would just throw something in, but then we would try it because we were so absolutely amazed and fascinated by the potential there, where I’m pretty sure that in some cases we overdid it because we got the shiny object. It’s like, “Oh, AWS is doing this. We probably should try it.”

But I think it was this also almost miraculous reality of the business needs and the advancements in technology really, really matching. But yes, it was AWS. It was a great time where right now I feel like we have so many choices that we spend a great deal of time even trying to figure out what the many choices we have. There aren’t that many, and AWS is pretty awesome. I’m a big fan of AWS as a technology organization.

David Joy:

The second part of what, just from this, I wanted to thread was you brought about, “Hey, we had this tendency to take a solution and maybe overfit that solution.” And we do the same thing with data too sometimes. But you brought up a very good point that for every company, you don’t really need to start by applying, over-architecting it. And we have seen some examples.

I don’t know if you know about this. Amazon Prime completely changed their architecture to serverless and then brought everything back to a monolithic architecture. They went-

Madalina Tanasie:

I knew that they were going to move to serverless and service-oriented architecture. I didn’t realize they went back, but I’m not surprised because even when you’re the creator of these patterns and you’ve demonstrated again and again that they are good patterns, I think you need to have that realization sometimes that sometimes simple patterns, even though your engineers might say, “Oh, they’re legacy and we are not modern,” really the right thing to do.

They’re going to find places and ways to be modern and innovate. But innovate in your space. Don’t innovate just for the sake of innovation.

David Joy:

Yeah. And I think the key is also, you’ve been in the space for a while, challenges keep changing. Scale changes, user requirements change. And back in the day, I don’t know if… I joke about this in pretty much every podcast is that something went down for a bit, if a server was down or something, max you would get was an email saying service was down. And “Yeah, we are working on it.”

Now, if something is down, you hear about it on Twitter. And the CEO is involved, and everybody’s like, “What happened to this service?” So, things have changed, paradigms have changed, and Collibra obviously has a pedigree in terms of what they’ve been doing. I was researching about the company, you have some great products around catalog and data governance.

But what are your challenges now, of course, from similar products? What are you solving at least priority-wise right now?

Madalina Tanasie:

Obviously, we have a large number of customers who are already using what we have. So, for us, it’s constantly about providing extra features in the areas that we are already pretty well-established. Scale. Obviously, there is always that one customer that is ramping up faster than we anticipated. And when people freak out about scale, I really like to reference Dr. Vogel who’s saying that scale breaks everything.

And instead of thinking it as a problem, you really just embrace it as just a measure of how successful you are. Because we all try to think about scale and plan, thinking like, “Okay, this is as large as this load is going to be or data set is going to be.” And you plan for that. And when you exceed it, it’s like, “Okay, we are more successful than we ever imagined that we’re going to in this area. This is good, people. Let’s figure out how we fix it.”

So, of course, as we grew and we got some of the really large companies out there use Collibra, we are really trying to figure out how we adjust to the tremendous amount of data. Metadata is a lot right now. We have just incomprehensible amount of data that these companies have. We only bring into our platform the metadata for that.

And we are speaking about billions of assets and billions of pieces of information that have to go there to have a full landscape of what the company has and where. And sometimes those objects are pretty complex. So, definitely trying to stay ahead of the needs of our customers is top of mind. But of course, AI has changed everything. And we are feeling pretty excited by how synergistic what we’ve built for managing data is with what’s needed to do proper AI governance.

And everybody’s speaking. I don’t know, I don’t remember us discussing a few months ago about AI governance. I think it was just a tiny topic there. But right now, everybody’s speaking about AI governance. And as we started thinking about what does it even mean AI governance, you realize that for every company, it means something else.

It ranges from what AWS and GCP and Microsoft are doing with really trying to understand the data observability and looking at those model security, access control, whatever, from companies that just want model repository and want to document all the use cases that they have because their customers or regulators are worried about AI being given up. So, it’s all over the place.

But the reality is that it’s both of these things. If you are to do AI governance, you need to understand your model and ensure that the model is for a specific given purpose and you constantly validate that the model does what you intended it to do. But there is also that part of model combined with data that you use for training.

And you need to understand what data you’re using, who’s changing the data, who has access to the data, how the data was transformed, and these are all components that Collibra has. So, we tremendously excited about the possibility to not only just be part of this AI wave, but meaningfully influence how AI governance is done and how the future is going to look.

David Joy:

And what you brought up is a very interesting point. For every enterprise, this is a different problem. And the way they look at this problem, especially what you said about looking at a RAG system where you have your own data and you have a large language model coming together and working. But you brought up a point that I was not thinking about that is, “Hey, people are changing this data.”

So, who changed what? Because what they change affects the overall result that the output will produce to the user. And it’s a fascinating problem to solve, and you guys are right in the middle of that.

Madalina Tanasie:

Because we’ve already solved that for different purposes, big companies and just companies that use Collibra have always wanted to know, for data sets that were particularly important for regulatory compliance or for business decision, they’ve always wanted to know who has access to the data, who changes it. And now, the same data products are used to train AI models.

It’s even more important if you’re going to use AI with all the anxiety and paranoia around AI and the unknowns around AI, it’s even more important to have those sensors and to do it right. So, I don’t think that we can do meaningful AI governance without really having the data governance in cataloging and lineage.

David Joy:

Yeah, that’s brilliant. You previously when we were talking about, and I’m going to take that quote as it is about scale that you said when you’re scaling and scale breaks everything. I had not heard that quote.

Madalina Tanasie:

Yeah, it’s not mine. It’s Dr. Vogel.

David Joy:

I’ll say I heard from you and then you quoted someone else. But that is a very profound thought, the thought that, hey, if something has breaks, and scale breaks everything. It’s also a good sign for you as a company to look and know that, “Hey, we are doing something right, that’s why this is breaking.” So, there is a positive outlook towards it.

How are you designed for scale at Collibra in the sense like your infrastructure? Is it running again on cloud? And how would you… tell me what you can without giving the secret sauce.

Madalina Tanasie:

It’s nothing that is necessarily a secret sauce per se. It’s very much about… philosophically, I think you need to prepare for success. But I really don’t think that you need to prepare for success that is so far off that you’re going to spend so much preparing for that success that you’re going to miss a train and you’re going to have nothing.

And I think this is a mistake that many startups do, especially now when there are all these patterns and everybody wants to be like Google. But you go and you overdesign your MVP, and by the time you have it out, you have three other companies that have done it and you’re out of business and out of money. And I think this can be the case at every stage.

You think what’s the next milestone and you design for that plus. And you have a plan about, “Okay, if we reach there, what’s our next step? How easy is for us to evolve and transform that?” But I think it’s also important to be very pragmatic and realize that nobody’s going to wait for you forever, that these companies have needs and they have them now.

And you have to constantly have that balance between speed and scalability. Don’t compromise on quality, don’t compromise on security, don’t compromise on reliability for the skill that you’re set to have. But in terms of scaling, there are trade-offs. And in terms of even uptime and reliability, there are trade-offs and you get to the point of diminishing returns.

So, I would say understand your business, understand what’s important for the customers, understand how fast they’re going to grow and how fast your fastest-growing customer is growing, and plan for the next year, the next two years, but don’t get crazy. And that’s what I’m trying to do as I’m working with the engineering team at Collibra, to anticipate what’s coming, but have that constant trade-off conversation between non-functionals and the need for features.

And the reality that right now there is this big need that everyone has is more important than anything else for us to-

David Joy:

For sure. Yeah. So, you have to be customer, user-driven, but at the same time engineer in the right way so that you’re not affecting overall balance of things. So, you kind of hinted a little bit, which I feel is also sort of like… I was curious to know was, and now as a leader, as a CTO, what are your engineering philosophy that you bring to Collibra?

Like you say to your teams, of course, you’ve mentioned some obviously, but are there others that you encourage the team to go after and do things in a certain way?

Madalina Tanasie:

I kind of hinted this. We have a very interesting mixed tech stack where we have cloud-enabled technologies. There was a bit of lift and shift that happened and then evolution. We continue to evolve. We have cloud native part of our platform. What I’m trying to do on the engineering side is really just have these guiding principles in terms of architecture and technology, and say, “Hey, we need to be as much as possible 12 factor because that’s super important to our ability to be reliable and to be consistent and to be able to move really fast.”

“We need to go towards building stateless products and solutions because that’s how we’re going to be able to auto-scale and that’s how we’re going to be able to have no downtime deployments and all these things that have become just able stakes as a SaaS provider.” And then, I really think that we should innovate in our space. If we can buy it, we should buy it. I don’t think we should reinvent the wheel.

We are multi-cloud, which is a complexity. At some point, we wanted to be cloud-agnostic and that meant that almost everything needed to be built by us, because you’re saying, “Hey, if I’m cloud-agnostic, I can deploy Collibra anywhere.” Well, there is nothing right now that I know of that works for everywhere. So then, we kind of said, “Hey, let’s take a step back and recognize that we are not cloud-agnostic. We are multi-cloud.”

There are these three, four big clouds, and we want to be able to be in those three clouds, so let’s leverage managed services. So, definitely very intentional buy versus build where possible so we can innovate in our space. And then, API first. No matter how beautiful you think your UI is, we are working in the enterprise space, there’s going to be a company out there that wants their own UI on top of Collibra.

And it shouldn’t be that we get into the business of writing 20 different versions about UI. Give them API, they build it however they want it. So, definitely, there are these guiding principles. But in terms of engineering culture, I think the team is more than the sum of the parts. And I think having people who really are committed to what you’re trying to do, who are fair players, who really work together, who are honest, who love freedom and are worthy of freedom, it’s very important part of the culture that I want to build.

David Joy:

Yeah. That’s awesome. You are the right CTO here. Look at you, mixing technology, experience and great quotes together. No, I’m really inspired by some of the things you’re saying, especially the idea of… when you brought up the idea that you have all these people working together and they need to be accountable to what you’re trying to build.

So, one of the things I was curious too was, engineers have this habit to go after certain things. They’re like, “Hey, can we do this technology? Can we do that technology?” So, there’s always this thought process of R&D and research. So, how do you encourage innovation, obviously, in R&D, but at the same time make sure that these are activities focused towards the product that you’re building or the feature that you’re building?

Madalina Tanasie:

I think there is a need for some space to play. I think engineers need that. And I think no one has monopoly on the best idea. The best idea, the breakthrough idea can come from anyone. And if you don’t create this time, play space, you might miss on really, really big things out there. That’s not to say that every idea and every prototype becomes a reality.

So, we have at Collibra something that we call Innovation Day. It’s something like a 10% variation of Google’s, 20% with a bit of structure around it. But we have these hacking days and sometimes it’s, “Okay, let’s fix the bugs, let’s fix our logs. Let’s fix whatever is the problem that everyone is most cranky about.” And then, there are some like, “Okay, let’s play with AI and let’s see what comes out of it.”

Or there is this thinking our product that it’s kind of… anyone has any idea. In terms of new technologies, again, spikes happen. There is this new thing you want to play, you want to try, you think it’s the right solution there. But we do have a process for introducing new technology because, again, it has to make sense. And you have to think more than just, “Is the right technology for solving this tiny problem?”

It’s like, “Okay, how are we going to deploy it? How are we going to support it? How often we need to patch it? Who’s going to keep an eye on it? Who’s going to be getting the alarm at 3 a.m., and all that?” And there has to be a really, really good need for us to introduce something and usually more than just one team needing it.

David Joy:

Yeah. I wanted to go back while you were saying this whole idea that you don’t want people to wake up. And again, you also have this problem of who the right person should be, who should wake up? And that stems from what you were previously mentioning, because you have to have a cloud-agnostic solution. That means you have to have somebody who understand GCP, somebody who understands AWS, and that’s a different problem.

We at Cockroach Labs, we have CockroachDB, which is also cloud-agnostic. And I’ve worked at another company before I came to Cockroach Labs, was also cloud-agnostic. And one of the challenges always was, “Hey, who’s the Microsoft guy who can help me with this problem that I’m seeing?” But you as a leader have to plan for situations like that because you have to have a balance of the right engineers, the right people. And hiring becomes a key aspect of that as well, right?

Madalina Tanasie:

Yeah. It takes a lot of discipline in making architectural choices to remember that you can’t just use everything that AWS has to offer because you need to have an equivalent solution in GCP and Azure. In some cases, working with the government, they want it on-prem. So, there are limitations in terms of what we can choose.

In terms of getting the call in the middle of the night, I don’t think we have that much problem that has to do with differences between clouds because we have uniform ways of deploying and monitoring and just run books that are applicable irrespective of what a customer or a service is hosted. Knock on wood, right now I’m going to jinx myself, we might have some weird problem that has to do with the clouding, but we haven’t had a lot of those.

David Joy:

What are you excited for as a CTO? At Collibra, what are you guys building? Of course, you mentioned AI. What can we anticipate? What can people who use AI at Collibra anticipate or just users of Collibra?

Madalina Tanasie:

As a technologist, more than CTO or CTO of Collibra, I’m really very excited about the developments in AI. And even as a parent, I think so much is going to be different in five years. How much has changed from December of 2022 until now? It’s unbelievable. A lot of the AI, and especially generative AI, was something that was for academics.

You had a bunch of PhD people looking and playing with it, and they were in their corner and nobody’s looking at what they were doing. And right now it’s everywhere. It’s every person, even non-technologies, they’re playing with it and they’re looking at it. And if you’re thinking about just how fast it’s adopted and how much we’ve democratized this ideation space and how many people can, in fact, leverage AI is amazing.

And I think engineering in five years, maybe less, is going to be very different. We already have Copilot that can code as a C5 engineer and can build test automation and do a whole bunch of stuff adequately. So, engineers need to learn how to do something else or how to use AI differently to differentiate themselves.

Medicine. You’re probably going to go to a doctor and the doctor is going to be a psychologist of source that is convincing you to follow your treatment plan and whatnot. Education. We already see with Khan Academy just how amazing is to have personalized education where that AI assistant knows where you’re weak and it just gives you the right lesson plan.

And so much is going to be different. And I find it fascinating as a technologist to see where we’re going to be in few years, given just how fast we are moving right now.

David Joy:

No, I’m with you on that. I feel like the speed is insane. I don’t know if you know this. Yesterday, OpenAI released a new model, which is insanely good. It’s called Sora, S-O-R-A. It does text to video. But it’s so good that it can… I showed it to a bunch of my friends. I say, “Hey, man, I shot this video. How do you feel about this?” And they were like, “Wow, this is really good.”

And then, I told them after five minutes, “It’s AI-generated.” And they were like, “What?” Insane. You should go check it out.

Madalina Tanasie:

No, I haven’t played with it. I feel like it’s one of the giants between Microsoft and Google. Each one of them kind of releasing new, trying to up themselves, which gets me to think “It’s going to be great. It’s going to be great for all of us.”

David Joy:

Sora is not available. It’s just videos that they had shared. I took that and shared it with somebody. But I’m with you on the fact that in the future, pretty much everybody that we know will have to adopt AI in some way in their workflow. Otherwise, they’re going to be a little bit behind in terms of, because the world is moving way quickly.

As you said, Copilot has become so important because you can code something that would take you six hours. Of course, you know how to code, but just AI Copilot really helps you finish that in an hour, one and a half hours, which is good for companies, a lot of things. So, it’s exciting.

Madalina Tanasie:

It’s more than that. Let’s say that you have a dependency on another team. Let’s say you’re a business analyst and you have a dependency of your data office and their data engineers. And you kind of go there and use your AI of choice, the GPT of choice, that your company has, and say, “Hey, can I have some Python code to do this?”

And before you know it, your dependency like, yeah, it’s not as good if a data engineer did it, but you’re on block. You’re not in a queue waiting for them forever.

David Joy:

True. That’s a very good point. So, I know a few questions, but I wanted to ask you one last question. What’s your advice to engineers who are on this path to leadership and CTO? What are the things that you’ve learned from your career?

Madalina Tanasie:

Obviously, I didn’t plan to be where I am, so I don’t know exactly if my path is a repeatable path. But I would say know why you’re doing it. Moving from engineer to manager requires different skillset. You have to let go of a lot of things. You have to do work through others. You have to build skills. You have to let go of some skills. It’s a big, big transition.

And you have to know why you’re doing it and for what reason. What’s your purpose? Is it selfish? Is it because you want to make more impact? All answers are valid, but the paths might be different depending on what’s your answer. What I would say to technologists right now is that just prepare yourself for a journey where you’re going to have to learn all the time. It was always the case.

Technologies are changing so much. I feel like I’ve been through so many technologies myself in the many years that I spend in software engineering, but right now it’s amazing just how much things are changing. So, you’re embarking as an individual contributor or as leader on a journey of continuous learning and you should be prepared for that.

And also what I said that was the biggest differentiator for me, truly recognize that as a leader, your job is more than just creating a technology vision and executing on that technology vision. It’s super important to explain it to the non-technical stakeholders in a way that they understand because that’s how you’re going to get buy-in and that’s how you’re going to make those great things happen.

David Joy:

Yeah. That’s amazing. Have you ever considered writing a book in the future?

Madalina Tanasie:

Yes. I always say that when I grow up, I want to be a writer for the New Yorker or something like that. And before I wanted to be an architect, I actually wanted to be a writer. Someday.

David Joy:

It’s fascinating to know that you’ve chosen to do what you do from your journey that began somewhere in Bucharest. And you’ve brought such a joy to me talking about this journey. And I’m super fascinated with what you and Collibra are going to do. So, thank you so much and we will catch you in the next one.

Madalina Tanasie:

Thank you for having me.