FinOps in Action

Why do so many FinOps practitioners miss the biggest savings opportunities, and what does getting technical have to do with it?

In this episode of FinOps in Action, I sit down with Nick Sudarikov, Finance Director of Technology Operations at InvestCloud, to break down how a background in finance and capital markets can actually make you a sharper, more effective FinOps practitioner.

Nick shares how he went from investment banker to hands on FinOps engineer, built a local AI powered stack to query billing data in plain English, and uncovered years of forgotten cloud spend hiding in plain sight.

We also cover why most cost tools will never catch your biggest waste, how to map cloud spend to real business value, and what it actually takes to have meaningful conversations with engineering teams. 

Here’s what we talked about:
  • Why FinOps practitioners should set up their own cloud accounts and get hands-on experience with the technology
  • How understanding engineering workflows changes the conversation with tech teams
  • The hidden cost problem that anomaly detection tools will never catch
  • How Nick uses DuckDB, Parquet files, and AI to query billing data in plain English
  • Why AI is set to boost FinOps productivity by a factor of 10 and which tools risk losing market share if they don't adapt
  • The importance of mapping cloud spend to business value before making any optimization recommendations 

Chapters:
00:00 - Cost Tool Reality Check
00:17 - Show Intro and Guest
01:36 - Why FinOps Needs Hands On
03:28 - Automation and AI Effort
04:23 - From Finance to Tech
06:57 - Early Cloud and Cost Lessons
10:58 - Startups vs Enterprise Scale
12:04 - Legacy Workloads and Risk
14:32 - Value Chain Before Optimizing
17:30 - Finding Hidden Waste Patterns
19:48 - Local AI for Billing Data
26:30 - Future AI and Closing

 🎧 Subscribe for more FinOps leadership conversations → https://www.finopsinaction.com/ 

Quote of the Show:
  • "If you haven't started a Docker container ever in your life, probably you're missing out on a lot of things in FinOps." - Nick Sudarikov
Links:

Ways to Tune In:

Creators and Guests

Host
Taylor Houck
Producer
Rayanne Pruitt

What is FinOps in Action?

Welcome to FinOps in Action! Join host, Taylor Houck, Each week, as he sits down with FinOps experts to explore the toughest challenges between FinOps and Engineering. This show is brought to you by PointFive - empowering teams to optimize cloud costs with deep detection and remediation tools that drive action.

FIA - Nick Sudarikov
===

Nick Sudarikov: [00:00:00] You find a great opportunity in the cost tool. You look at the numbers and say, Hey, we're spending so much on this. But then you try to unfold that like, how that situation actually started and what are the challenges, and so on. And you understand that it's much more complex than that.

Intro: welcome to FinOps in Action. I'm your host, Taylor Houck. Each week I'll sit down with FinOps experts to explore the toughest challenges between FinOps and engineering. This show is brought to you by 0.5, empowering teams to optimize cloud costs with deep detection and remediation tools that actually drive action.

Taylor Houck: Hello, and welcome to another episode of FinOps in Action. Today's guest calls himself a refugee investment banker, turned FinOps practitioner. He started his career in venture capital and capital markets before making his way into the cloud. First at Priceline, and now as finance Director of Technology Operations at InvestCloud, [00:01:00] a FinTech platform with over $6 trillion in assets running on its infrastructure.

But don't let his finance background fool you. This guy is technical. He recently built a local AI powered stack using open source models to query billing data in plain English. The way that he thinks about FinOps is deeply rooted in engineering best practices. Welcome to the show, Nick Sudarikov.

Nick Sudarikov: Hi Taylor. Great to meet you.

Taylor Houck: Super, super excited to be chatting with you, Nick. And I think that, uh, the audience is gonna have a lot to learn from you. When you think back on your, your experience in FinOps, right? Having transitioned from, You know, finance and capital markets into technology. What are, what are some learnings that you've had over those years and where do you think that FinOps practitioners commonly go Wrong?

Nick Sudarikov: Well, I think if you haven't started a Docker container ever in your life, probably you're missing out on a lot of things in [00:02:00] FinOps.

Taylor Houck: deep on that, Nick, because I think you're, you're gonna rub some FinOps practitioners the wrong way. I would reckon that probably fewer. Then 25%, maybe fewer than 10% of FinOps practitioners have actually done that. Wh what do you mean, and, and why do you feel that way?

Nick Sudarikov: well, I think it has to do a lot with the way we explore tools, the way we understand software, because if you start with just thinking of the FinOps framework. You're kind of bound to certain rules and you don't have that creativity with the tools. Just, You know, a lot of tools look vastly different when you log in as a user, where you log in as an admin, like you see much more. So my advice to everyone doing FinOps is to set up private accounts with cloud providers. Like set up your own org is actually free for the first 12 months and try to. Look around, like how do you start a vm? What services are present there? How do you analyze those services? That helps a lot. And [00:03:00] You know, through my career with Priceline and now with InvestCloud, uh, a lot of the optimizations were driven by that exploration and learning, like what are the actual work loads.

Taylor Houck: There's something to be said too with the deeper that you understand what the tools are that your engineers are working with, and the closer that you can relate to them and their challenges and their priorities, the more effective you will be as a FinOps practitioner.

Nick Sudarikov: Absolutely. Yeah. You know, when we do any optimizations work and most tools in the market would kinda rate that work for you, they would say whether it's slow effort or medium effort or high effort, but You know. Those qualifications, they're always arbitrary. Like for someone, it could be low effort for someone, it could be high effort.

And now with ai, uh, with ai it's changing like resizing 300 VMs. That could be high effort for some, like if you are using legacy workflows within your tech org. But if [00:04:00] you're using AI or just a basic script, I mean you don't really need intelligence for that. You just need scripted workflow. Uh, that instantly becomes very low effort. So. Understanding the platforms you work with, understanding engineering workflows? Uh, I think it's a huge asset for anyone in FinOps and that is really transformative because that changes the conversation.

Taylor Houck: take me back to when you were first transitioning from finance into technology, because I think that a lot of folks, when they hear your background, they actually wouldn't expect you to come out talking about using Docker yourself as a a user. What, what was it like when you first made that transition and how did you prepare yourself to go deep into technology and engineering?

Nick Sudarikov: I was always passionate about technology. Like I was always curious and I wanted to become a software engineer from. Somewhere around high school, then I figured that, uh, I lived in Russia at the time. I'm originally from Russia. uh, all the [00:05:00] education around software engineering is centered around math.

Like first you have to do crazy math without numbers, just Greek ladders for like four or five years. And only then you actually transition to doing something meaningful, something you can play with on computer And so on. And just standing in front of the chalkboard. Got me tired really fast, so I opted for economics and finance. But I've always been around computers, like, You know, I've been 11 or 12 years old. I would res solder some stuff in my motherboard, like do all the upgrades myself, write some simple code and basic, that was fun. That was fun. And I kept on doing that throughout my career. Like whenever I was doing anything in spreadsheets or working in any documents, I was always trying, uh, to find ways to automate stuff or do something more efficiently. touch

Taylor Houck: It, it's almost like you were a technologist, like [00:06:00] masquerading as a finance guy for a couple years, and then you, you came back home to, to engineering and, and finance.

Nick Sudarikov: You know, a lot of things in finance and, uh, I did finance for the technology sector, you really have to understand what you're doing. So, uh, previously I worked for a company that originated in Russia and then operated across the globe. Uh, I took it public in London in October, 2021. And through seven years with that company, we did a lot of tech M&A. So we had to really dig deep into. What these companies do, how their operations are structured, like if they're selling software, if they're integrating software, what they're doing with it, how we can improve their operations And so on, and operate in that mode. You very often have to go an extra mile, like you have to look deeply into the products to understand where value is created, how it is created, and how you get more out of it.

Taylor Houck: Tell me a little bit about your, your [00:07:00] early days in the cloud. And when was it when you realized that the cost of your cloud infrastructure was something that you wanted to focus on in your career?

Nick Sudarikov: You know, the company I used to work with, uh, it was a major cloud reseller in Europe, so I wasn't that much concern about the cloud cost actually on the country. We wanted to, wanted our customers to spend more on the cloud because that's how we earned commissions and, uh, being one of the major partners with Microsoft back then, You know, I think around 20. 18, 20 19 large tax started seeing that transition for resellers. So they didn't want to pay resellers for just renewing enterprise agreements or like basic software licenses because it's not recurring revenue for them. And, You know, back to the world of finance, uh, a RR, which. stands for annual recurring revenue?

Is that what everyone tracks? I mean, that's basically the performance indicator for a company. And around that time I saw [00:08:00] that transition with major software vendors. So they start tracking that cloud was a natural focus. So we wanted to sell more. We wanted to sell more cloud services. We wanted to sell more consulting services that naturally. Help customers transition from their on-prem environments to cloud And so on. That's when I discovered that and started exploring, You know, we were supporting a lot of startups, uh, in, uh, Eastern Europe. So we were partnering with Google with AWS to provision startup accounts for them, like essentially operating those platforms, for some of the venture capital projects I work with and. I started exploring, like I set up my own accounts with providers, like see what I can do, how do I start stuff? Like can I run my VM there to connect to something I can to. Like watch video elsewhere. You know, sometimes people use VPN, but uh, whenever you use a publicly [00:09:00] available VPN, you might not be able to connect to some of the services, like let, like Netflix.

But if you set up a private server with a dedicated ip, uh, those locks don't work And so on. that's how I discovered that. And I also learned that AWS egress traffic is very

Taylor Houck: Yeah, that's It's hard to realize that the, uh, the decisions you make on your cloud architecture can actually have a big impact on what you're paying at the end of the month. I'm curious, when you think back to those early days when you were working with folks on transitioning from the data center into the cloud, was cost a primary driver of that for them or was it mostly for the technology and then the cost was just, uh, You know, something that came with it?

Nick Sudarikov: I would say it was mostly for the technology and for ability to scale. know, even now when. People think of on-prem, usually if you just calculate units, in most cases that would be cheaper. I mean, if you run it smart, that would be cheaper. There is private cloud. the problem is, uh, when you start to scale, [00:10:00] you constantly have to over-provision and the variation in your workloads can just not be, uh, might just not be sustainable in the on-prem model. And. At the same time, if you are not sure about the future workloads like you, you have a project You know will be running for six months or so, it does make sense to build a million or billion, do dollar data center for that,

Taylor Houck: Especially for, I mean, you were mentioning you were working with a lot of startups that were just getting started. I mean, they have all the runway in the world. They could be growing 10 XA year and then need more infrastructure than they ever could have imagined, or the opposite could happen. Right? And they need to scale down and the cloud gives them the flexibility to do that.

But with that flexibility comes opex, right? An opex that is intrinsically, uh, usage based, right? And then when you start getting the bill every single month and the bill becomes a larger percentage of your overall opex spend, especially with an engineering orgs, it becomes something you really want to focus on.

But now, shifting gears a [00:11:00] little bit, Nick, from there you went on to work for a, a very large enterprise, helping them to manage their cloud spend. What were the key. Differences when you moved upstream into a massive enterprise operating, You know, very large scale cloud deployments from when you were working at an MSP with many smaller companies.

Nick Sudarikov: When you work for a large enterprise. that actually runs workloads that resells those to customers And so on. And, You know, I was not in an engineering role with my previous job, so I learned that there are a lot of variables in those workloads. Uh, there is a lot of politics internally, like how you do stuff And so on. So you can't just take the. Numbers approach and look at how much things cost because there are a lot of dependencies. Like, You know, some workloads that seem inefficient are actually the backbone of the entire operation.

And you can't just take them [00:12:00] down and re-architect them or, so you deal with a lot of legacy stuff.

Taylor Houck: I mean, this is so important to recognize, and it kind of ties back to your original point, is you need to understand the technology before you come in and start recommending changes. Because, and I've talked about this before, but I, I previously worked at N BBC Universal, helping them with FinOps, and a lot of our spend was on legacy SAP workloads running in the cloud.

Now you can look at the utilization rates and tell them, You know, this looks highly inefficient. You gotta do something about it. But then under the covers you recognize, I mean, number one, it's a critical corporate application. That is a, You know, if that thing goes down, you've got more problems than the cost of the infrastructure.

And then also you have limited ability to make changes because you must run it on supported infrastructure. And the last thing is that the engineers that are responsible for the infrastructure are the same ones that are gonna get paged at three in the morning. If there's an issue. So if you just willy-nilly make changes as the FinOps guy, Hey, [00:13:00] I'm good.

It's Friday afternoon, I'm checking out for the weekend, they're gonna be called if there's an issue.

Nick Sudarikov: You nailed it.

Taylor Houck: How do you, like what, what did you recognize? Like how did you come across concepts like that when you were at, uh, at Priceline?

Nick Sudarikov: It's experience. I mean, you start, you find a great opportunity in the cost tool. You look at the numbers and say, Hey, we're spending so much on this. but then you try to unfold that like. How that situation actually started and what are the challenges And so on. And you understand that it's much more complex than that. And You know, for someone in FinOps, uh, just dealing with part of the overall company, spend a certain number they see, uh, might seem huge, but you don't, you, you don't always know like the revenue that's behind that and how this process affects the businesses, the business of the company. So. Looking at larger numbers or probably escalating that to senior management, you start to understand that things are slightly [00:14:00] different and it might be worth wasting a couple of thousand, a couple of hundred thousand dollars in some service, just not to take any risk associated with that process.

Taylor Houck: I mean, a hundred thousand dollars per month seems astronomical to us in our normal life, right? When we're like filling up our tank of gas, it's like 50 bucks and it feels expensive. But then you recognize like at this enterprise scale, a hundred thousand a month could be cheap. It could be nothing, right?

Nick Sudarikov: for many large businesses could be a million a month or more.

Taylor Houck: here, here's something I want to pick your brain on, because on one hand, yeah, a hundred thousand dollars a month is nothing. A million dollars a month is nothing. But that doesn't mean that you should just spend it and not worry about the efficiency of these workloads.

And the reality is that. Cloud infrastructure is complex, and these billing models are complex, and any given resource has many configurations that you could place on it. And the most optimal configuration from a [00:15:00] cost perspective is highly dependent on its utilization, right? So how do you bridge this gap between, hey, this is a critical workload and we need it.

So a hundred thousand dollars or a million dollars a month is nothing and. Yeah, that doesn't mean that we should just be wasting money on, You know, inefficient architecture or misconfigurations.

Nick Sudarikov: There are two, uh, items to this question. So first is that before, like right now, my framework, before I start any optimizations or even asking questions like how I plan before I start planning my questions to tech teams And so on, I try to understand the value chain of the products, like how our. Tech process, how our tech org aligns with the value creation within the company. are the processes that we run, how those contribute to. The revenue. And after that, I look at those processes and whether something is inefficient or if or underutilized because You know, in [00:16:00] FinOps sometimes provisioning more resources, especially as we deal with some serverless items, you provision more resources, the job gets done faster and you end up paying less. But from the beginning, that could have looked totally different.

Taylor Houck: when you talk about tying the engineering resources and cloud architecture to business value, can you think of any? Um, situations or, or stories or anecdotes about that to help the listeners kind of put it into practice and, and bring it to life?

Nick Sudarikov: I'm not sure what I should, like, how detailed I should go on this one because I signed some NDAs And so on. But basically within any tech org you would. Probably have different parts of your products. Like you would have a piece of operations that generates data. You would have piece of operations that stores data. You would have some processes that feed that data to customers. You would have some external connections to products And so on. So to me, that was really important to understand, [00:17:00] how product utilizes each technology stack that we have. How those modules interact with each other. I think that's crucial.

And You know, in most businesses, I mean any large scale business, you would never understand that to the fine detail and then definitely not in your first year on the job, but at least getting some helicopter view of that definitely helps. Like without it, it's very hard to start and it's very hard to have meaningful conversations with tech teams.

Taylor Houck: All right, Nick, I'm gonna shift gears, uh, a little bit now because one of the things that's really interesting for me when I chat with many fit ops practitioners is that a lot of us have, uh, some stories about when we got hooked into this thin ops thing, and usually it has to do with driving some large, uh, optimization effort that had a big dollar amount impact.

When I say that, any, any stories or or situations come to mind around, uh, optimization and, and saving money?

Nick Sudarikov: That was early in my FinOps days, [00:18:00] I was excited because I got read only access to one of our accounts. I think that was AWS uh, it didn't have. I'm not sure it had a storage lens enabled. That's when you can actually analyze what your buckets are, and so on. I was trying to understand like what I can do here, and there were no anomalies in the cost tool that we used.

So I just started looking at the names of the folders, like some names of the buckets, and I found a bunch of stuff that has like 2016, 2017 in it.

Taylor Houck: What year was it at the time?

Nick Sudarikov: I think that was. 2023 or 2024. So that was like well over five years old. And after that, I started looking into cost tools, like identifying those items and what I have, like, You know, most cost tools, they track anomalies, but they don't track long-term patterns.

It's not a anomaly, it's been there since 2016 it it's the same.

It's [00:19:00] not an, an anomaly. So yeah. So whenever I analyze, spend. I would always isolate that at the resource level. Like basically I would go through all resources and build a chart, say over last year or over the last three years, like whatever you got. And if it goes like this, if it just goes up, might have a problem there.

I mean, you might have a super successful product that's always growing, like compute is, uh, growing, you need more storage And so on. But. Quite often that's not the case. And especially if you see flat, um, if you don't see any growth in other resources, like if you have a single resource that keeps growing, growing, growing, then you might have a problem there. So I, I, I would do that manually for some resources. Now I can use AI for that. That's super helpful.

Taylor Houck: let's, let's dive into ai because I know from speaking to you that you've been doing some interesting things with local AI development. Can you just give me a high level overview of some of [00:20:00] the AI projects that you've been working on recently, especially as they relate to FinOps?

Nick Sudarikov: absolutely. Yeah. So the first project I did with AI is I get really tired of watching my cats on camera. So I, I have cameras at home and. I, I can see my cats when I'm not there. The problem is that very often nothing interesting happening there. And I, I, I just, I just want to watch my cats moving.

Like I'm not interested in lights changing And so on. So I was thinking like, can I have some tool that would actually, track that for me? Because, uh, whenever you use motion detection that's built in camera, it's not precise.

Taylor Houck: If the sun like goes down, the shadow will move.

Nick Sudarikov: Exactly. Yeah.

They, they don't understand the actual event that's going on camera. Start thinking about that. And then I learned that, uh, you can actually use a vision model with it. would first process, uh, the frames it gets from camera [00:21:00] and detect motion for you and provide a description of what's going on. the cat is sitting on the counter And so on. So, uh, naturally I wanted to play with that. Uh, I could install a smaller model on my Mac, uh, get the API endpoint for that point, the video recording software that I use for that to that endpoint, and it start providing descriptions of what's going on. And then it's entirely open source like. talking about that.

Like you can get Ola for model, you can get frigate, that's, uh, the video surveillance thing And so on. I mean, that's incredible what you can do locally these days. I wouldn't imagine doing anything like that five years ago.

Taylor Houck: Right. And it almost feels like the acceleration of these AI tools is happening so quickly that there's been a pretty major shift in the utilization of these different. Models and, and tools even since like, let's call it December of 2025, I'm curious how, as it relates to [00:22:00] FinOps, how you're, how you're leveraging these, these tools.

Because watching cats is cool, don't get me wrong. I also, my, my first project that I built using, using Claude was just something to clean up my, my personal email inbox, because I had like 8,000 emails in there and I built a tool that could help me scrape and categorize and clean it up. But then you get into like real work use cases.

How are, how are you using it day-to-day to help, You know, scale you as a FinOps practitioner?

Nick Sudarikov: So whenever you use a cost tool. comes with some limitations. It's getting slightly better as MCP servers are rolled out across tools, but sometimes just your struggle with interface or you wanna layer multiple parameters to get the idea of usage And so on. And it's just much easier to ask questions in plain language. what I did, uh. I started analyzing, uh, detailed billing data with ai, for that, uh, I use a very simple stack. So you can just download [00:23:00] Parquet files or, uh, zip files with your billing data. Uh, yeah, Kerr, essentially Kerr. Uh, I prefer Parquet files because they are much smaller than, uh, zipped, uh, CSVs like by magnitude.

And they're also much faster to work with. So I installed local Doug db and then it, uh, essentially becomes sql. And the cool thing about that is that I can use any tool like or Chat GPT or claude or whatever to write queries

Yeah. And essentially, uh, like. The billing structure for all the cloud

providers, it's public. So you can just paste the schema into your ai, uh, chat bot and ask questions in natural language, then debug. Then if you see didn't find something, you start changing those queries manually adding like, like instead of just a fixed expressions expression And so on. And it helps a lot. You know, you can do so [00:24:00] much more exploration with an approach like that, uh, compared to standard cost

Taylor Houck: Yeah, I'm, I'm curious because, You know, I genuinely want to pick your brain on this because I've got a lot of thoughts about where this thing could go, but if you extrapolate it from here, right, obviously you've got all of your billing data sitting there, and you can create or generate SQL queries and grab whatever tables and grab the data that you want and analyze it.

As it comes to visualizing this, right? A lot of what FinOps does is cost reporting, right? Showback, chargeback, cost dashboards drill into this. Drill into that. Have you started playing around with visualizing your data using these AI tools?

Nick Sudarikov: I mostly visualize my data with cost tools or just put it back into PowerPoint or Excel and visualize it there. I mean, that would be fun though.

Taylor Houck: My vision is that you can generate, you can generate pretty, [00:25:00] um, advanced like visual elements using just HTML and CSS, right? So if you have AI tools that can generate HTML and CSS, uh, in, let's call it a, a very. Easy, cheap, and easy way. Then any of the queries that you go and run using your AI tool, you could then pass on to like this visualizer agent that would then put it into a visual, You know, tool and then you could host it somewhere and like even a serverless web app on Lambda, and then pass a shoot, shoot someone a link, and they just open the link and, and there's the, the dashboard or the analysis.

Nick Sudarikov: That sounds really cool, but I think it's important to stay focused.

So your. Like with this podcast and everything, you're transitioning to content creation territory. When you do thin ups, you're. only have 2% of time for content creation. So you have to focus on [00:26:00] analysis and the way you visualize data.

I mean, it's great if something looks cool, when you do that in an enterprise setting, like, You know, uploading something so it's available as an online document or something like that, that's a security nightmare. So I, I, I think it's important to use tools for the right purpose and whenever you can do that looks great.

I mean, I like to play with those things on my side projects. But we still have to do our jobs during the day right?

Taylor Houck: That's absolutely true. So where do you see. These tools being helpful, let's, let's put ourselves maybe a year in the future. Right. Where do you see AI changing FinOps practices within major enterprises?

Nick Sudarikov: I think it is a much broader conversation. So it's not how AI changes FinOps practices, it's how AI changes enterprise software. And it also depends on where the tools go.

So if tools embrace this transition, they set up integrations, [00:27:00] MCP servers And so on. I think that would just give us more ways to consume data and explore it better, find more optimization opportunities. If some tools don't integrate with ai, they might be losing market share. Because to me, building a dashboard somewhere that's not connected to AI might take a few hours, like pulling in all the data And so on. whenever you start working with ai, it essentially boosts your productivity by a factor of 10 or sometimes even more.

Taylor Houck: So

Just to put a bow on on this conversation, Nick, if you were giving advice to a FinOps practitioner, let's say they've been doing FinOps for. Five years. Right. They haven't gotten on the AI train besides just, You know, asking general questions to chat GBT as if it was Google. Right. What advice would you give to them to start using [00:28:00] AI in their FinOps work

Nick Sudarikov: I think the best you can do with AI is, uh, data analysis and data mining. where it's most useful.

Taylor Houck: how would you reckon someone get started? Along that journey

Nick Sudarikov: Ask Chad, GPT.

Taylor Houck: that is. Absolutely. Awesome. Um, Nick, this has been, this has been so much fun. I've really enjoyed learning from you. I think the audience has also gotten a, a lot to learn from you. But before I, I let you go, I wanna make sure we also get the opportunity, opportunity to learn, uh, about you a little bit. I know you mentioned earlier you were born in Russia, I believe you've lived in Moscow, in Stockholm and in New York.

How does. How do you think about, You know, your international journey and how that plays in and also working in different companies across different countries. How are, uh, how's it different to work in companies depending on where, uh, they're based or where you're based?

Nick Sudarikov: Work ethics is country specific, [00:29:00] so I would say. Compared to Europe, US is a bit, it's much more results focused so people are ready to work longer hours or work more productively. That definitely helps.

Taylor Houck: How do you, um, in, in your experience, because I, I actually had a conversation earlier with a phenomenal practitioner who was based outta Europe. He was really focused on, on sustainability in his practice, and actually was talking to me about how working with his engineers. The more that he put sustainability and carbon impact on the front of like the front page and like cost was almost behind that he saw more success.

I'm curious if you, um, have any experience, uh, with sustainability, um, and now it relates to FinOps.

Nick Sudarikov: I have some experience with sustainability. When I lived in Sweden, I had to sort everything. was hard. took a lot of time, but I, I loved it. That was kind of fun.

Taylor Houck: What do you mean by that sort? Everything.

Nick Sudarikov: Yeah. Like before [00:30:00] you throw anything out, like you sort plastic, you wash plastic, And so on. And it takes some time, and I was always wondering like how it ends up. Like I, I wanted to hope that it's not going to landfill anyway.

Taylor Houck: I hope not, man.

Nick Sudarikov: Yeah. I, I, I know a lot of companies in Europe, uh, now have, uh, regulatory requirements to report on carbon footprint And so on. I just think it really depends on the type of business you operate. Probably if you are hyperscaler, then it makes a lot of sense. The space we operate in, mostly consuming cloud. I don't think, uh, that we really have a lot of direct impact onto carbon footprint. The important thing to say here is that if you're running outdated workloads, like older versions of Python or in generally running inefficiently, you're definitely not contributing to carbon footprint reduction

goals.

Taylor Houck: I mean, the way that I've thought about [00:31:00] it, and it's not a direct like one-to-one comparison, but generally speaking, the lower cost your workload is, it's probably related to how much power it's consuming. Right. So if you make a more cost optimal decision, it probably means you're consuming less power.

As power is a big, You know, input to the cost drivers of these hyperscalers, right? So it's like you kind of can get a directional sense, but tracking it down, uh, like the, the data is actually quite hard to, to find. And anyone that's measuring carbon impact in, in FinOps is it's directional. Um, at best.

Nick Sudarikov: And that could be translating to a lot of day-to-day decisions, like what model do we want to use for this AI workload? Like do we really need super fine precision? Or we can go with a smaller model that would be much cheaper to run and that would have a lower carbon footprint.

Taylor Houck: Well, Nick, this has been an absolutely amazing episode. I, I really do think that you're one of the most forward looking, um, practitioners that I've had the pleasure of meeting. Uh, I'm curious, this is one parting note, um, to the listeners. Where, where [00:32:00] are you consuming your AI news and staying on top of the latest and and greatest innovations?

Is there a, a book, a podcast, a resource that you leverage and would recommend to, uh, to the audience?

Nick Sudarikov: A little bit

Yeah.

I, I Google, I ask AI tools, uh, like nothing specific. You know, great article might pop up everywhere.

Taylor Houck: Uh, well on that note, Nick, if people want to get in touch with you, where's the best place to find you?

Nick Sudarikov: Uh, you can find me on LinkedIn or Slack.

Taylor Houck: Well, Nick, this has been a fantastic episode. Thank you so much for coming on the show.

Nick Sudarikov: Thank you,

Taylor Houck: you soto our audience. If you got something outta today's conversation, which I'm sure you did, please share this episode with someone who needs to hear it. This has been another amazing episode of FinOps in Action, and I'll see you next time.

Outro: That wraps up another episode of Fit Ops in Action. Thank you for joining. For show notes and more, please visit fit ops in action.com. This show is brought to you by 0.5, [00:33:00] empowering teams to optimize cloud costs with deep detection remediation tools that actually drive action.