Back to podcast library
The Data Hustle transcript

Building Data Platforms From Nigeria, Luck Surface Area and AI's Hidden Costs — Joy Victor

Joy Victor, data platform engineer at Federated Engineers, on owning a data platform end to end, selling a thankless job to the business, increasing your luck surface area to win international clients, and the hidden cost of AI data centers in Georgia.

Episode 19 · Nov 13, 2026Transcript length: 43:48Download transcript
Building Data Platforms From Nigeria, Luck Surface Area and AI's Hidden Costs — Joy Victor thumbnail
Episode 19
Full transcript

Read the full episode transcript

Generated from the exported YouTube subtitle track. Timestamps appear roughly once per minute and link back to that point on YouTube.

0:01
Tony and Juan

You are listening to the Data Hustle podcast, where data is turned into business impact. Joy, thank you for joining the podcast today. So for those listening in, what motivated your move from research data analyst to data platform engineer?

0:23
Joy Victor

Thank so much for having me, Tony and Juan. It's a pleasure to be on this podcast. so my move from research data analyst to data platform engineer was strongly motivated by ownership. I wanted to own a lot more things end to end. I first started as a research data analyst, where a lot of my work was around using data to make sense of a lot of happenings in the world. So, for example, I got to work with, you know, amazing people in Georgia, specifically with the AIAI Network in Georgia, and that was a beautiful experience where in all what we did, we saw how data was impacting people. In car insurance, we saw that people living within specific areas got to pay more than others, and we got to see how racism was kind of like an underlying factor in why they were paying a lot more.

1:23
Joy Victor

And another thing I got to do was looking into security, how AI is affecting security. You could say I was ahead of my time with that particular project because I and my team were able to predict a lot of things that are currently happening with AI, basically making attacks a lot more insane and the scale it's at right now. I was able to present my work in, Rwanda and just talking about how other African countries should prepare, as well as how Georgia and Atlanta should prepare as well. Most recently, I worked with data centers looking at how-- basically researching deeply into how AI and data centers themselves are affecting communities. Looking at the cooling system, the electricity, water consumption, all those, you know, nerdy stuff basically.

2:13
Joy Victor

And my transition to data platform engineering, like I said, was, you know, motivated by ownership. I've been in the research and academic space for a very long time, and I was like, "What really goes on in the industry?" You know, "Let's, let's go find out what happens with those bros over there." And I wanted it to come from a place of ownership because it's-- I've seen how AI is moving and how data, again, is what makes AI what it is. The more data you have, the better. not, not necessarily, but usually the data systems largely determine how your models are going to perform. And I was like, "Okay, you know what? I'd like to own the system end to end."

2:55
Joy Victor

Everything starting from ingestion all the to how the data is used. And that was what, you know, spons- that basically ownership was what motivated me. And since then it's been really, really hard work because no one told me that like when you say, "Oh, I want to do hard things," you actually get to do hard things. I should have known, but yeah. and that was so interesting for me because it was like a hybrid of multiple jobs. So at-- you're a data engineer, at the same time you're sort of like a platform and DevOps engineer, and somehow, somehow you also have to bring in your analytical skills as well. And it's, it's a very, very big role. It requires you to know a lot of things about networking, security, infrastructure as code, distributed systems, how data works in the real world, how it moves, you know, backfilling, so many things, right?

3:50
Joy Victor

And one of the hardest things I think I've had to deal with is the fact that like there's no, you know, one magical word course that you can just watch and like, "Yay! All your answers... You have all the answers, all your problems are solved. Get your certificate, go look for a job." that didn't work. So I had to learn multiple, multiple things. I think one course I had to check-- Juan has a couple of videos on YouTube about Airflow and... Not Airflow, sorry, dbt, and it was one of the most helpful videos I've ever watched. Shout out to you, Juan. And, it was just-- It's actually quite interesting to see that you get to learn from multiple places.

4:30
Joy Victor

You're just picking things and doing things. And I've worked on like absolutely exciting projects, fascinating projects with companies that I absolutely love. And yeah, that's where we are now.

4:41
Tony and Juan

There's so much fascinating stuff to get into there, and I'm so to see you're a fellow data platform engineer too. maybe Joy just a quick question. Can you give us an example of some sort of ownership that you were lacking back when you were in academia?

4:59
Joy Victor

Okay. so one of my most recent projects, I built a data platform from scratch. So I worked on everything from how the data comes in. So the data comes in, you know, in an S3 bucket, and then we had to take it to Snowflake so the analysts can work on the data in Snowflake. So starting from-- Like, when you think about that originally from top of your head, it's like, that shouldn't be so hard, right? They should just, you know, write dbt models, you know, transform the data, get what they want," da da da da. But what happens under the hood is that how does this thing work, right? So the first-- one of the first things that comes that you have to do is how do we manage this as code?

5:41
Joy Victor

That's easily one of the first things. We want to make sure that every dbt model is managed as code, even the Snowflake infrastructure managed as code, so we know who's changing what and at what point they're changing things. We also need to make sure that, security is also very important, so I had to implement role-based access controls. the engineers cannot have access to-- The data engineers have certain things they have access to. The analyst obviously shouldn't have access to everything. And So the, the platform basically looked something like this. So you push a dbt model to GitHub and then triggers the CI/CD pipeline. I had to build the CI/CD pipeline.

6:26
Joy Victor

Basically lints everything, checks everything to make sure that everything works really well. Then once that's done, it packages your model and pushes it to ECR, and the ECR basically manages it as an image. And now you have Airflow. Airflow runs in Kubernetes, and Airflow actually is making sure that once there's a new image in the ECR, the ECS basically kind of like triggers the ECS. And once... Oh my gosh, it sounds like a lot of things are moving So with the ECS, you have ECS going on there, and ECS basically picks up that image, spins a container from that image, and then you now see that your Snowflake, basically your schema or your tables and everything is transformed.

7:15
Joy Victor

So previously, what I would do is just, you know, do research about certain, industries and see how those industries work. But now I'm actually making sure those industries are working. And there's a lot of behind the scenes work that goes in that nobody sees. And, to be fair, I always say data engineering or data platform engineering is a very thankless job. Like, nobody sees what you're doing really a lot of time. But what you're doing is very important to how everything, you know, moves on in the future. So for example, like someone seeing this platform that I said I built, right? Might be like, "Why are there so many moving parts? Why so many things going on?" But then again, you don't want someone to push wrong code.

7:58
Joy Victor

So the CI/CD is there to ensure that bad things don't happen basically in production. And then we also have, you know, the role-based access control in Snowflake to ensure that people don't have access to too much things. we don't want a situation where, say, bad actors have access to credentials of the analyst. We want to ensure that, okay, they only have access to a certain amount of data. They don't have access to everything. So you try to limit as much as possible what people have access to. So in a way, you're kind of like also security conscious as well while building your, your platform. And so, so many things you have to put into consideration.

8:38
Joy Victor

And usually nobody notices the good work you've done until something bad happens. And that's when like, "Oh my gosh, you saved us. Thank you so much. You're our hero." Yeah. Or sometimes you're on call like, "Okay, this thing broke. This person's dbt model is making everything, go haywire. You need to roll back." And because the platform is set up in this way, you can easily roll back. So that's pretty much it.

9:04
Tony and Juan

And among data consumers, I feel that they get used to like, oh, this is normally how a data platform is, and every company has exactly the same one. Yeah. But it's only as I go from company to company and hear the stories that we realize, like, whoa, you guys have an amazing platform here.

9:21
Joy Victor

Yeah. Yeah, you really don't know until you, you move around and you see the way some peop- some companies run their things and you're like, "Oh my gosh, this is... This doesn't look nice." Yeah.

9:34
Tony and Juan

Yeah, I mean, I guess, I guess you're giving us the, the, the recipe to get recognition. Just leave in small things that break and fix them magically in five seconds. That's it. Yeah. So I, I wanted to ask, I mean, you've worked on so many interesting things here that are kind of at the forefront of the field, but I, I think what is more interesting to me is that you're doing this, you know, abroad. I mean, many people that we speak with will generally work in the country that they live or even in, in the continent that they live. But you've worked, like, you know, projects across the globe, I mean, in Europe and in the US. I'm kind of wondering, like, what is the process of, of, of getting there, right?

10:18
Tony and Juan

I mean, there's gonna be a lot of people in your position that, also wanna do the same thing, and they're wondering like, "How do I, how do I get opportunities like that?" Do you think it's, like, technical skill, how you present yourself online, or relationships, or something else?

10:34
Joy Victor

I think it's a combination of many things, to be honest. First of all, I, I would say it's a mindset thing. You, you start with your mind and, one thing I strongly believe is that you can do great things from small places. So even... 'Cause I live in Nigeria, and not just in Nigeria, I don't live in one of the big cities. I live in a very small city in Nigeria. So getting people to trust you and, getting to work with international companies, clients is, you know, a big ask, right? Especially when it's with data. Okay. So, so there's multiple things. It's a combination of both, you know, having good technical skills because you must be very good.

11:18
Joy Victor

And a lot of times, I'm usually maybe the only African in a room and, I have to do my best to make sure I represent myself and my people. 'Cause you don't want a situation where you are substandard just be- and they're like, she's African, that's why." you don't want to give people that excuse, so you have to be very, very, very good at what you do. So for example, I did a generative AI in data analysis course that was sponsored by the University of Miami and Knight Foundation. It was great, but like I was the only Black person, among the instructors, and I had to show up so well because like I needed to do my research, I needed to You know, work extra, extra hard to make-- 'cause I wasn't just representing myself now, but it was people.

12:06
Joy Victor

But in seeking opportunities like that, you have to be kind of like public-facing a bit because people can vouch for you if they kind of like know you. And there are people I've worked with for years that I've never seen them, like physically in the flesh, just Zoom calls and Google Meets, and we have a great working relationship. So you have to be a bit more public-facing, so you have to probably like tweet more, put yourself out there on LinkedIn. But then that's not all. You also need to network and have good relationships with people. You need to, always make sure that you're build- because usually it's someone that tells someone, that tells someone, that tells someone, and they're like, "Hey, I think I have this opportunity for you."

12:50
Joy Victor

So it's a combination of all three. You have have very solid technical skills. You need to, you know, talk more in public about what you're doing. Otherwise, these people are millions of miles away. They probably will never see you ordinarily. They probably will never hear of you ordinarily. And you also have to make sure you're also building a solid network of people, 'cause again, people know people and people talk about people.

13:12
Tony and Juan

that's absolutely right. It's like your network is your net worth, if they say, right?

13:17
Joy Victor

Yeah.

13:18
Tony and Juan

I, I, I can really relate to a lot of the things that you're saying because for a lot of my life I've been the only Latino in a room. And so like, to me it's, it's always been a- about like making sure that it stops being like the Colombian guy and that they like, "Oh no, that's Juan, and Juan is his own person, his own work, his..."

13:37
Joy Victor

Yeah. Yeah. And there's also another thing of also like helping people come up as well. So making sure that like you don't stay the only African in the room for long. also trying to help your own people and support them, giving them opportunities as well. So instead of say, okay, you get an offer and like, "Oh, I, I, I don't have the availability, I can't do this." Like, "Oh, I don't have the availability, but I know someone who does." And then you are, you know, giving other people opportunities as well. yeah.

14:11
Tony and Juan

Oh, interesting. That's one I haven't tried though. yeah, we-- it is interesting what you're saying, right? That's the, the standard for some people is higher than others, which is, something I think, not everybody is willing to admit, but it's the truth.

14:29
Joy Victor

It is the

14:30
Tony and Juan

I, I think it's very good to make that explicit. I mean, y-you know, if, if we're talking about network, I mean, y-you did a really good job, right? Like, you're in a small place, as you say, out in the countryside. So, you know, for people wanting to, like, network, what would you give them? What, what, what sort of practical advice would you give them to, to, to do this?

14:53
Joy Victor

the number one advice usually I would say is join online events, 'cause you can't be there physically a lot of the times. So join online events a lot. And don't just join and be quiet. Yap, right? I think probably from the few minutes we've had on this podcast, you'll notice that I'm a yapper. so yap a lot, right? Put in-- put comments, right? You see someone's YouTube video, you really like it, go drop a comment there. you see someone on LinkedIn, TikTok, drop comments, and it's just a matter of time they will get to notice you, who you are. same thing with joining events. When you're joining events, ask questions, right? You might be participating virtually, but ask questions.

15:39
Joy Victor

Be active. When you're done, you can create a post on it and say that this is what you're learning and all. As much as possible, you should try to increase your visibility. And, there's something called, that I've, I've seen a lot of people talk about it. They call it your luck surface area. So try to increase your luck surface area. The, because luck is a very interesting thing. You can't exactly like, you know, tell it when or how to come, but you can manufacture as many opportunities for yourself as possible. So many ways for people to, to find you. So for example, you found me 'cause of my videos, right? You didn't know I exist previously.

16:20
Joy Victor

So if you can, if you're comfortable in front of the camera, you can do that. If you're not, you can write. I used to write a lot. I only started making videos this year. I used to write a lot more. So you can write, you can yap on social media if you can. Just as much as possible, increase your luck surface area. Because again, these people don't know you. They are miles and miles away, and this is like your only chance of them ever coming across you. And another thing, yes, before I leave that out, is build really cool stuff. I think building really cool stuff also gets you a lot of visibility. 'Cause if you build something that solves a very serious problem or you build something that is really, really fun, people get to see it, they use it, and then it increases your visibility, increases your opportunity, the amount of opportunities you get.

17:15
Joy Victor

So I think those two.

17:18
Tony and Juan

that is, that's really interesting. I mean, this is kind of what we're seeing, you know, whether people are looking for jobs, or they're a consultant, or they're running a business. Really, what it all boils down to is you create a lot of public assets for people to get excited about and to get to know you, and you give them... You make it easier for them to know you exist and to engage with you, right? Like, that is essentially what you're saying here. Yeah.

17:42
Joy Victor

Exactly.

17:44
Tony and Juan

Yeah, it's, it's always a funny feeling when somebody approaches you out of nowhere, like in a conference or an event like, "Hey, I know you from this." It's like, "Oh, cool. I forgot that existed."

17:54
Joy Victor

very interesting. Yeah.

17:56
Tony and Juan

Yeah. Right. to, to, to, to rewind a little bit. So you, you were talking about, data platform engineering being a thankless job. So- How do you start this discussion, right? Like, with a stakeholder or a client, how do you make them see the value without it having to require an outage, right? 'Cause if you're doing your job well, they're never gonna see something going wrong and you coming in to fix it. So in those situations, you might be doing such a good job that they think you're not

18:26
Joy Victor

Doing

18:27
Tony and Juan

something. So what-- How, how do you, how do you approach that?

18:33
Joy Victor

I think first of all is as much as possible, I'm not sure how much of that is within your control, but try to work for someone that appreciates data. Let's start there because working with someone that doesn't appreciate the value of data would frustrate you so I can't even go into details, but that will frustrate you 'cause the person doesn't see the value in what you're doing to begin with. But, I think one of the things to always do is to explain what you do. As much as possible, get people to buy in to what you do. So during stakeholders meetings, you get to explain the value of what you do. So for example, you took this pen from here to here, right?

19:20
Joy Victor

Just something simple, right? But don't just say, "Oh, I took this pen from here to here." Instead say, "I took this pen from here to here. This is the problem it solves. This is how much more efficient we've gotten as, as a result of this. This is why I think we should have done this. This is why I took the initiative to do this." So you always have to give them things they can see, feel, or touch. So things like money. Money would always get stake- stakeholders any day, anytime. So explain to them that like, "Okay, this thing that we're doing is saving us X, Y, Z amount of money," or, "This thing that we're doing is costing us so much money and let, you know, roll back on this."

20:00
Joy Victor

Or things like AI, for example. Everybody has, you know, all the AI bugs. Everyone wants AI, but they fail to realize that if you don't have good data, you have, you know, mid AI. And now you start explaining the situation to them. So if, if you want us to have, you know, an AI product, here's, here's the reason why you need your data infrastructure to be solid. Because one, imagine if it gives wrong advice to a customer, chances are you're going to get yourself involved in a lawsuit. Nobody wants that. If you give your, agent, for example, root access, because, you know, AI can do pretty much anything right now. But if you give your AI agent root access to provision infrastructure in production, these are the things that could happen.

20:46
Joy Victor

you could have personal, information leaked. You could... So many things could go wrong. Production DB could be dropped, no backup. So many things could go wrong. You explain to them that these are the things that are likely to happen, so let's not do that. Or these are the things that are likely to happen, so let's do this. And say, for example, if it's a company that's just focused on analytics dash, but you have to explain to them that if you have the wrong data, if you don't build the structure properly, we're going to make bad decisions, and these bad decisions would affect the company. We would lose money, we lose customer trust, we would lose, so many things that we stand to lose or so many things that we also stand to gain.

21:27
Joy Victor

We can make better decisions as a result of, you know, having good data. We can actually be able to catch issues quickly before they arise publicly. So you just have to connect it to something that they can relate to because it's very abstract. A lot of what we do is abstract and, again, a very thankless job. Nobody wants to do it. It's dirty work, to be honest. But once you connect to things that matter to them, once you connect it to, numbers, trust, people, and all of that, they start to see the value in it.

22:04
Tony and Juan

I, I use a lot of car analogies. It's like a platform, kind of like a highway, and an analyst is a driver, and this would be like a seatbelt. The moment-- I, I'm not even that big into cars, that's the funny part. But the moment that I relate like data engineering concepts into cars, people are like, "Okay, yeah, that makes sense. You need seatbelts."

22:26
Joy Victor

Yep. Yep. Exactly.

22:29
Tony and Juan

Juan, I, I, I would equate it to, you know, you need your engine to run well, but you don't know how it works. It's a black box. Yeah. Yeah.

22:37
Joy Victor

Also true.

22:40
Tony and Juan

Governance is like traffic norms.

22:46
Joy Victor

Nobody likes obeying traffic regulations, but no one wants a ticket either, so you always have to hit them where it hurts.

22:55
Tony and Juan

Well, talking about like making things more abstract, one of the, the questions that we have here that I thought was very interesting to ask is: what do you think are things that people are misunderstanding about data engineering today?

23:11
Joy Victor

I think that's a great question. so what would I say a lot of people are misunderstanding about data engineering? Pretty simple. Everybody says data engineering is moving data from point A point B. Sounds simple theory. It's not that simple in reality 'cause you're moving it across like multiple places and now all of a sudden you have to start thinking of costs, right? If you're moving it across like different availability zones or across different like softwares, different providers and all of that, you now have to start thinking about how much that costs. No one told you that at the beginning. And now you, you started concerning yourself with financial operations.

23:57
Joy Victor

And, another thing is networking, right? If say you have Airbyte running in a Kubernetes cluster, all of a sudden you need to know networking. You need to know how Kubernetes itself works, and Kubernetes is also like something very complex as well. So, so many things are like moving and nobody tells you. Another thing I mentioned earlier was security Because data is such... It's pretty much gold, right? And you have to protect it as much as you can. Otherwise, again, the company gets in trouble. Everybody's, you know, there's a security breach, something happens. So you also have to protect it. All of a sudden you are a sort of, you know, cybersecurity analyst of some sort.

24:38
Joy Victor

And so it's just, it's just not one thing anymore. Data engineering is no longer one thing. I think we are gradually moving beyond, just writing SQL queries, just writing simple Python scripts. We're gradually moving from that into doing like more complex things and solving more complex problems. Infrastructure as code is there. You need to know how to use multiple cloud platforms. there's also, you monitoring. Monitoring is grad- gradually like entering the scene, so you need to start knowing how to use things like Pulumi and stuff like monitoring. Like you- no one prepared me for this, okay? So, but all of a sudden it's now gradually part of your, of your job description.

25:22
Joy Victor

So again, data is no longer moving it from point A to point B. It's how much is it costing me to move data from point A to point B? How secure is it to move data from point A to point B? how long will it take to move data from point A to point B? If along the way getting data from point A to point B, it gets stuck, what do I do? what happens? If the data did not move from point A to point B on a regular basis, how do I fix that? You know, and if any data was, you know, outdated as a result, how do we fix that? And how do we use this data to make a decision? So it's not, it's not as simple as people put it. There's a lot more work that, is involved in the entire process.

26:09
Joy Victor

So that's why I think a lot of people misunderstand about

26:12
Tony and Juan

the data world is full of these, very simple ideas that are very difficult to implement. I can think of like self-service, single version of the truth, moving data from A to B. Even data is the new oil, but like, okay, but how do you monetize data? A lot of these things are very simple in theory, but like the practice is where it gets very difficult.

26:36
Joy Victor

Yes. Yeah, it is. It's, it's always easier in theory, like getting into a production environment, yo, it's a lot harder. Yeah.

26:55
Tony and Juan

Also, sometimes when I see vacancies and I hear people speaking about data engineering, I see two very distinct profiles. One of them is like, querying data with, I don't know, Polars / pandas, knowing SQL, and it's very like SQL-heavy, analytical-heavy. But then there are people that are just a hundred percent platform engineering, so lots of cloud, lots of Terraform. Like, what are your thoughts on that? That there are some data engineers that have no idea even what infrastructure as, as code is or why is it important, but then you also have like some very hardcore platform people that don't know SQL.

27:37
Joy Victor

Yeah. I think I was having this conversation with Tony, before we started recording, and it was... What we talked about was AI engineering specifically, and how like the roles these days pretty much say anything and everything. So for example, you can have an AI engineer who only knows how to use automation tools and that's it, and they're an AI engineer because that's what the job description calls them. And then you have an AI engineer, which is pretty much a data engineer that feeds data to AI models or machine learning models, and they are also an AI engineer. And then you have an AI engineer that actually builds these models, and they're still an AI engineer, but they're-- the- all three AI engineers do entire completely different things, right?

28:26
Joy Victor

So I think it's, more or less industry nomenclature. So we say, "Okay, this is a data engineer and this is what data engineers should do." And, then there's also what I've noticed as the rise of, data platform engineers as well. So we're now looking for data engineers that know a lot more about platforming, infrastructure and all of that. We've also seen, software engineers where we see like the job title is software engineer comma data. So they're looking for a data engineer or someone that understands distributed systems, how, you know, microservices work. Like they need you to have like a lot of things together. The long and short of the story is that I think a lot of things have to do with like what the job description is requiring.

29:13
Joy Victor

But my personal opinion these days is that I think people... AI can pretty much write a SQL query right now and write Python scripts, dbt models and everything. So as much as possible, a lot of people now are trying to upskill, so they get to do the complex things that AI doesn't know how to do. And that is where we're now seeing people, data engineers knowing more of infrastructure as code, data engineers knowing how to work with both, streaming data and batch data, and knowing how to work with, monitoring systems or Kubernetes or Docker and stuff, containerization, all of that. So it's, it's kind of like necessity is giving forward the rise of innovation where people are trying to learn more and do more.

30:07
Joy Victor

So yeah, that's pretty much it.

30:10
Tony and Juan

That's true, you're really good at...

30:12
Joy Victor

though.

30:14
Tony and Juan

Now, if you're really good at platform, everything data platform, that makes you like very, I'm gonna say unique, because the amount of people that have the specialty to have, like Joy would say, full ownership of a data platform It's really small. So that's, I also can imagine one of those jobs where people might be more okay, like looking further than just like which people are in my own city.

30:43
Joy Victor

And it kind of like gives you an edge in the job market because the job market is quite brutal right now. So the more skills you have, the better your edge basically right now.

30:57
Tony and Juan

So maybe what are those-- like what skills would you recommend people to learn today that make their career like very strong?

31:07
Joy Victor

AI.

31:12
Tony and Juan

I should have known. Yes. You made it too easy, Juan. yeah. That's it. Done.

31:20
Joy Victor

yeah, once you know AI, you know everything. You're fine.

31:22
Tony and Juan

Yeah.

31:23
Joy Victor

yeah, I'm joking. So, but to an extent now, I've noticed that a lot of companies are requiring you to, at the very least, know how to use AI tools. So just the basics, you know, know how to use AI tools to improve your productivity. And in some cases where you'd be working with models specifically, you'd probably need to know how to use AI to interact... How to use like your data to interact with those models. So a bit of knowledge there is like usually great. But if you want to go into engineering today or data platform engineering today, what I would advise is have a solid understanding of your basics first. And by basic, I mean Linux. Let's start there 'cause I'm actually very surprised at how like overlooked Linux is.

32:12
Joy Victor

Probably because maybe I'm on the platform side of, of data engineering. You get to use Linux every single day, like every single day, and it is one of the least mentioned things in the data engineering journey. So start with Linux, know how to move around with your environment and just move things around. 'Cause a lot of times things will not work on your computer, right? the models are not running on your computer. They're most likely running in a server somewhere. And if you probably, for whatever reason, have to interact with that server, you need Linux. So, and there's no graphic user interface going on there. You are going into the dark ages to work with that.

32:54
Joy Victor

So Linux, very good. Python, fantastic. SQL, very good. Then a cloud provider as well. That could be AWS, Azure, GCP, depending on what is like very common in your environment. Then you could also look into things like, infrastructure as code tools. So basically you can now provision cloud infrastructure because you've learnt about cloud providers. So now you're using things like Terraform to now provision infrastructure, and now you have to like move data from place to place. So look into Airbyte, Airflow, and to transform the data, dbt is like the industry favorite. There's so many things. Oh my gosh. There's so many things to learn. What am I missing?

33:45
Joy Victor

You probably need to know how to use like different data warehouses. Data warehouses, data lakes. depending on the providers, you need to learn Snowflake. again, that is like very dependent on what you're interested in. You could just, you know, stick with AWS and, you know, everything AWS provides or you can explore so that you can be provider agnostic. but usually it's better start with a single provider that you're comfortable with or a provider that can give you as much free credit as possible to start, because it's expensive to learn. But one thing I would definitely recommend, like I would definitely add is give yourself grace and enough time, 'cause I, I don't think three months...

34:31
Joy Victor

'Cause a lot of courses and bootcamps and everything usually says three months. If you are coming from zero experience, that you have no idea what Linux is, you have no idea what SQL is, you have no idea what Python is, three months is not enough Especially if you want to be really good at what you want to do, three months is not enough time. So give yourself enough grace to learn as much as possible and build stuff on your own. Break stuff on your own. Not in prod, on your own. And then that's where, you know, all the learning comes in and stuff. So you... There's, there's so much to learn, to be honest. And another thing I usually tell people who are learning is you would most likely never see that one YouTube video that connects everything you want.

35:20
Joy Victor

So say for example, you want to learn how to use Airbyte with, maybe a simple database. Chances are you're going to find 10 tutorials on it, and those tutorials are using different kinds of databases. So it's always going to be a very interesting journey. Very, very interesting journey. So you always have to very open to change, always have to be very open and flexible with learning. Yeah. All I'm going to say is it's not easy. It's really not easy and there's a lot to learn, and you should just give yourself grace and a lot of time.

36:02
Tony and Juan

Yeah, that makes a lot of sense. I mean, talking about curiosity and learning things. I, I kinda wanna, get back to the topic that you're working on, which is, like, the data centers and the environmental impact. What is something surprising that you found working on this, the problem space? 'Cause I think everyone is kind of wondering now, like, what's gonna happen to the environment and the data centers we're building? It's all, like, super fast, like, way faster than anybody has had time to consider the consequences. So what, what, what are you seeing?

36:32
Joy Victor

So the interesting thing about data centers is that they've been in existence for a while, but AI, you know, ballooned the whole situation 'cause where have you been storing all your photos, right? Cloud storage. It's in a data center somewhere. but now with AI, we need a lot more, a whole, whole, whole, whole lot more. And that's why it has now become an issue of concern because... And one thing a lot of people don't know is that you do have data centers in industrial area. Like, not just in rural areas, like, you know, on the countryside where nobody lives. You do have some in regular office buildings. A lot of people don't know that. So the data centers are regular office buildings, and it consumes a ton of electricity.

37:26
Joy Victor

And what's fascinating about it is that the government or a lot of governments in the United States, basically... Okay, I'll speak for where I did my research. It was specifically Georgia-focused. What happens there is the government is actually subsidizing for a lot of states, in a lot of states in America, subsidizing the electricity cost for data centers. So data centers are paying less for electricity, meanwhile, regular residents are paying a lot more electricity. Some are also consuming water without getting permits. They're consuming water. They are releasing harmful chemicals to the environment without permits. So that is where everybody's like, you know, "Okay, you are polluting our environment.

38:20
Joy Victor

Our water pressure is getting worse, and we are paying more for electricity." There's literally no benefit to the individuals that live in such areas. So now, we are now basically bringing more to these things because I like to say that AI is the train that has left the station. It's not coming back. Like, no one is going to the, you know, probably stone ages of not knowing what ChatGPT was. I don't think anyone is doing that. If anything, I feel like everything keeps getting better and better and better and more complicated every single day. Like, I think last week there were like three releases. Meta had its own release. Everybody's releasing something.

39:03
Joy Victor

I'm not sure who's, who saw Sam Altman and, like, Elon Musk fighting on Twitter. That was fun. I had a good time. It was so much fun. but what we do, what we are doing is ensuring that people are aware of the dangers behind these things. And for example, in Georgia, when... Okay, so let me just backtrack a bit. Data centers need cooling a lot because you are having, you know, a lot of heat is being generated, and so it needs to cool. And there are different ways cooling is managed. Some waste more water than others, but there's absolutely none that doesn't include the wastage of water. They all do. And now when this water is no longer needed, which is one of the major problems that no one looks into, when the water is no longer needed in the system, it is expelled.

40:02
Joy Victor

And that water is no longer fresh. It now contains chemicals, and it is now, you know, put into probably a wastewater facility for it to get treated, and it's not treated properly in some cases because this is new technology. These are new ways of things happening, and the wastewater facilities are either insufficient. So, for example, in the Fulton County area of Georgia, they only have six wastewater facilities, and in contrast, they have over 50 data centers. So imagine six water facilities, wastewater management facilities managing those, you know, that large amount of data centers. So there's a lot of things that could go wrong and that are going wrong.

40:50
Joy Victor

People are reporting environmental health concerns. People are reporting about their plants dying, fishes in the area dying and stuff. And as much as we know how capitalist the society we live in is We should also raise awareness so something can be done to mitigate the situation. Because again, we only have one planet, and as much as possible, we should do our best to save and protect it. As much as we're getting benefits from AI, we should also make sure that we stay alive to enjoy those benefits. So that's pretty much what we're doing. And yeah, there's a paper on it. I can probably share a link to the paper so everyone can have a look.

41:33
Tony and Juan

Of course. So yeah, maybe if you have a data center close to your hometown, get a water filter just in case.

41:41
Joy Victor

Yeah, you should. You should, absolutely.

41:44
Tony and Juan

Didn't think of that, but, I will check, see where they are in my house. I, I will have to do some, research here on my end as well, actually. Thanks for, raising awareness to that. Yeah.

41:57
Joy Victor

Yeah. You're welcome.

41:59
Tony and Juan

Okay. So, Joy, what is next? Where can people find you?

42:05
Joy Victor

Okay. I think I'm on Twitter. I talk a lot on Twitter. people say I rant a lot, but to be fair, my work is very demanding, so I need an outlet. So you'd find me on Twitter. it's J-H-O-I-E double underscore on Twitter, and that's pretty much how you find me almost anywhere else. On LinkedIn, it's Joy Victor, and I create memes on TikTok, I find fun. So on TikTok, I am thejoyvictor on TikTok. You can also find me on, I'm also like a very community-heavy person, which is also one of the ways I actually do recommend people learn. so if you want to learn, if you want to gain more opportunities, join a community. Always join a community. Have people like you that could, that can usually help make your journey faster.

42:56
Joy Victor

So I have a-- I'm a program manager at the Data Engineering Community, so you'll find me there. I'm usually helping people, supporting them, running programs, bringing guests like yourself. If you're probably interested one day, you could show up.

43:10
Tony and Juan

Sure, let me know. Sure, yeah. We would love to.

43:14
Joy Victor

Yeah. I'm also, a data platform engineer at Federated Engineers. Federated Engineers is a data consulting and training company, so you can also find me there as well.

43:28
Tony and Juan

Awesome. We'll make sure that everybody can find all of these things in the description when we release the episode. And, thank you again for joining us. We-- This was a really interesting and fascinating conversation.

43:39
Joy Victor

Yeah, it was fun. I really enjoyed it. I loved the fact that it was just like talking to my friends. Like it was so

43:46
Tony and Juan

we appreciate it. Nice.