FinOps in Action

FIA - Chandan Bukkapatnam
===

Chandan Bukkapatnam: [00:00:00] just broke into it from there and I realized how it was really an addictive career, uh, being involved so deeply intertwined with both cloud, um, and seeing the bigger picture of the finance por portions of it. Here's something that it is actually a focus of ours that we have to actually spend time on, not just the engineering and operational excellence, but also the financial and cost optimization excellence.

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, I am really excited to be joined by someone with a unique path in our industry. is still a relatively young discipline. of us transitioned into it from engineering infrastructure [00:01:00] or even finance. But today's guest started his career in FinOps. Coming out of a computer science background, he intentionally chose to build his career at the intersection of cloud architecture and cost optimization. the past several years, he's honed his craft in the discipline and has built a reputation for driving meaningful impact. He's now bringing that expertise into his role as a senior Cloud FinOps analyst at Procore Technologies, Chandan Bukkapatnam, welcome to the show.

Chandan Bukkapatnam: Hey Taylor. It's great to be here, man.

Taylor Houck: No. Absolutely. I, I'm so happy to have you on the show and excited to dig in. Um, I mean, I guess we can start with where the interest started.

Chandan Bukkapatnam: Mm-hmm.

Taylor Houck: your career in FinOps. How did you even know this was a thing,

Chandan Bukkapatnam: Yeah, I mean, I, I'd like to probably start off back to, to the college days. Um, as, as a computer science student, I, like, I kind of got into it, um, after switching my major from Aeros, um, or sorry, astrophysics, and I kind of [00:02:00] was just like. I don't, I don't wanna do astrophysics. I'll go to like the next best thing, which is computer science.

And then about three years into it, I was like, I don't like coding. I don't think I wanna be a software engineer. So, um, and, and at that point I was kind of pretty, uh, deep into the world of finance with, with stock trading and, and futures and all of that good stuff. So I was just kind of looking through career options for stuff that kind of gets into the intersection of computer science tech.

In finance and You know, like investment banking, that, all that good stuff. Um, and just kind of through my search, I came across the FinOps organization, which was where I initially started. Um, I did not take a whole lot of classes with cloud computing or, or. Like just cloud architecture as a whole. So that was a new one for me as well.

And looking into the organization, seeing how there was a middle ground between tech and finance that I had previously not known about, that was a really cool thing for me to kind of just like. Dive into a little [00:03:00] bit, look into some of the certifications. Um, and from there I was just kind of looking at FinOps roles.

Um, and at that point I, FinOps really wasn't a role that I saw on LinkedIn or Indeed or any of those, those places. Um, I was just searching for like cloud optimization, FinOps. Didn't really come across a lot, um, except for one position at Hertz, which was where I initially started my career off. Um, and like you had mentioned, even in that the, the role requirements were like five plus years of experience in either engineering or finance.

And, and I was like, I don't have either of those. So, um, I, I just kind of applied on a whim and, um, I got, um, I got a chance to interview and, and I, uh, I took it and, and it went well. Um, and kind of just broke into it from there and I realized how it was really an addictive career, uh, being involved so deeply intertwined with both cloud, um, and seeing the bigger picture of the finance por portions of it.

It was, it was really cool to, to start my career [00:04:00] off there. And, um, yeah, I haven't had any reason to switch out of it, man. I've, I've, I've loved it ever since. And then I'm kind of just excited to keep going down this path.

Taylor Houck: I think that it was, was a genius move for you to look and identify a field like this that's growing so rapidly and also combines the. The disciplines of like, so many different aspects of business, whereas like when you're coming in out of, out of school, you, You know, you can You know, this, these things across so many different layers without the baggage of, of any experience that you had.

I'm curious, when you, when you came into your role and got started, what were your initial like goals and, and, and how did you think about, um, the, the role and the job and, and the impact that you were able to drive?

Chandan Bukkapatnam: Yeah, no, that's a good question. I think like, first off, I just wanted to like set myself some context as to like what, what we're trying to achieve. Um, and initially at that point I had obviously heard of AWS heard of the cloud, but I wasn't. [00:05:00] Really knee deep into it. Like I didn't know what a M four, like four X large was.

I didn't know what any of these terms were. So, um, I initially knew that this role was a little bit more technical facing rather than finance. So, um, I knew that I had to go in and just study, like learn the different services, um, at least get like a basic understanding of what. All the different services were the cost implications of them, um, as well as how the company currently sits and, and the current infrastructure of the company.

Um, and just understand from people, like what, what are your challenges today on both the technical side and the finance side? A lot of the stuff you hear with FinOps on the finance side is, is visibility and understanding what. Any of this stuff really means like, what does it mean to be overprovision?

What does it mean to be like saving, having a bunch of savings plans that are underutilized, like that context for finance wasn't there. And then on the technical side, um, coming from, um, a company that was very legacy heavy, um. They weren't really [00:06:00] all that, all that focused on usage based metrics. They were kind of just like, we need the same exact infrastructure that we had on-prem, on cloud.

So, um, if the utilization is lower, it's okay because the stuff is running well. So understanding and giving them the context on the finance side and giving finance the context on engineering side, that's, that was the biggest initial challenge for me. Um, and yeah, I mean, just the sheer amount of services that AWS offers, having at least a basic understanding of all of that.

That was, that was at least a, a three to six month endeavor right there. I mean, getting those certifications and doing all that stuff, that was, it was rough initially, but um, now it's kind of just second nature.

Taylor Houck: Well now, I mean, you're several years in, it's like we talked how you started your career in FinOps. Well, FinOps has been around for a little while now. You have several years of experience under your belt at this point. When you look back at those, those early days or look back at, You know, your experience, generally speaking, when it comes to FinOps, what is something that you think that, I mean, perhaps [00:07:00] you got wrong or could have improved upon?

Like what, what do you, what do you think about.

Chandan Bukkapatnam: Yeah, I think I focused way too heavily on cost optimization at a super granular level. So I'm talking like. Looking at EBS volumes that would save the company $3 a month. And, and looking at five of those and pestering engineers to go take care of those, like, um, I spent so much time making together an Excel sheet on all these little optimization opportunities.

Because I didn't really understand the total dollar figure yet, and I didn't understand the levels of impact that that takes or that requires of the engineers. Obviously, like for EBS, we know that it's, it's a low implementation effort of just deleting these volumes, getting a snapshot and just forgetting about it, but.

When you think about it from an engineering perspective, this is an additional lift that the engineers have to do. Um, and for, for them to focus on five actions that would save the company 15 bucks a month, that's just not a good use of their time. And I spent so much time putting together a list of all [00:08:00] these opportunities that when you looked at the total sum, it looked like four or five grand.

It looked like so much money at the time to me, because I'm like, that's a lot of money. Um, but you're not thinking about it holistically from like. The company spends X, Y, Z, like F four to five grand is not a whole lot of money. Uh, obviously these need to be taken care of, but when you ask an engineer to do 50 different actions to save 4 40, 500 bucks a month, that just doesn't make a whole lot of sense.

So when I initially started off, I was super focused on the tiny pieces of it. Instead of trying to figure out a system around how we actually solve these, not just as a one-time activity, but long-term and make sure that we don't actually have to worry about these things over and over again.

Taylor Houck: So how did you, how did you do that once you made that? Realization. Hey, these little onesie twosies, they're not really moving the needle and perhaps, I mean, it may not even be worth. engineering effort that it takes to resolve these things. I mean, one of the numbers that you don't see in your spreadsheet is the hourly rate.

You know, that you're paying for these engineers. I mean, fully [00:09:00] baked in. Let's, let's include, You know, everything. And it, and it's gonna, not every little optimization is worth it, but. That doesn't mean that there's not a lot of opportunity to leverage their time to save money in the cloud. How do you move from, You know, the, the individual small opportunities that are $15 a month into a program that systematically helps you to measure and optimize your cloud spend?

Chandan Bukkapatnam: Yeah, I mean, for me, what I, what I initially wanted to do there was, I mean, I, I obviously knew opt, uh, automation existed, but I didn't know how to go about implementing it. So I wanted to initially figure out like, how do I make this worth the engineer's time and. I can't ask them to do like 50 to a hundred different actions every couple weeks.

So what if I ask them to spend maybe 20 to 30 hours figuring out an optimization or automation workflow so that they can then actually just do it as a one time and it gets taken care of automatically. [00:10:00] So that's kind of where my head went after like three or four months of just no success with getting any of these small optimization opportunities taken care of.

So. Obviously we, we leveraged our AWS teams, our, our support systems. Um, and we kind of came across the compute optimizer at that point. And, um, we, AWS is slightly working on a little bit of an automation engine within Compute Optimizer then, but, um, obviously you, we kind of stuck with our Lambdas grips.

Uh, we stuck with some of those kind of, uh, those, those easier low hanging fruits. Um, and we just went with some of those smaller EBS ones. So unattached EBS volumes after 30 days. Delete them. Like that was, that was a super simple automation script that ended up like resulting in a lot of money saved at that point.

Um, and that kind of proved it like, as a proof of concept there, that automation works. Um, doing the, having the engineers work on this for 20 hours one time rather than having them do this like every other week makes [00:11:00] so much more sense. So we kind of took it a little bit more of that way where. Um, a lot of these smaller items just got taken care of by automation, but we focused on the bigger ticket items like the, the stuff that resolved or related to architecture or like some misconfigurations and, and spending our time on stuff that was actually big ticket and letting the automation take care of the smaller ones that don't really need a lot of human intervention.

Taylor Houck: Yeah, I mean, it, it, it's so important to put those guardrails in place and they can be so successful. I mean, I think back too. In my early days of, of FinOps, um, I was working at MicroStrategy on their development team and we, we implemented this automation that pretty much after every sprint we would just purge all dev, EC2

Chandan Bukkapatnam: Oh yeah.

Taylor Houck: Unless they had a tag on it that was an exclusion. Right. So it's kind of an opt out rather than an

Chandan Bukkapatnam: Okay.

Taylor Houck: And it was something that when you think back, I mean this was like 2019, it was actually a little bit. Uh, advance for, for

Chandan Bukkapatnam: Oh yeah. Yeah.

Taylor Houck: Kind of seems, it [00:12:00] kind of seems obvious though, that, that it's something that, that you should do and automate and that then it, it frees you up from just focusing on, okay, let's, let's look at our dev instances.

And that's a problem that's gonna keep coming up and keep coming up. 'cause you have, You know, we were doing two week sprints, so it's like, what are you gonna every two weeks go look manually through the entire list and then reach out to people, Hey, what about this? What about that, that you consider now that's taken care of?

Now we can move on to. The next thing to

Chandan Bukkapatnam: Mm-hmm. Exactly. Yeah. I mean, but even, even with that, like, I, like you said, that's, that's an extremely advanced way of thinking and um, the fact that all the engineers and like the engineering leaders were able to get on board with that, that's a big piece there too. Like obviously dev is not something that you necessarily need 24 7, but.

It's, it's also one of those places of contention where people don't want to turn things off. People don't want to delete things. That's a scary feeling to, to be in charge of deleting that. So, I mean, kudos to you guys 'cause that that's, that's an insanely advanced move that [00:13:00] early.

Taylor Houck: Yeah, I mean, I will say that we, we, the, the, the context behind here is that cost optimization was a very high priority from the very senior levels of our, our leadership.

Chandan Bukkapatnam: Mm-hmm.

Taylor Houck: was, it was a, a, a top down decision. But I'm curious, kind of, kind of getting into that, how, how do you think about. Um, actually making things happen, right?

Because as you mentioned earlier, you can come up with recommendations, you can have conversations, but at the end of the day, FinOps doesn't have most of the time the authority to like actually take an action.

Chandan Bukkapatnam: Yeah.

Taylor Houck: How do you make sure that that happens?

Chandan Bukkapatnam: No, I think that's twofold. Um, initially, like the biggest piece for me is getting an engineering or finance leadership as like the stakeholder, the primary stakeholder for all of FinOps. So somebody that we necessarily don't report to, but somebody who's really intertwined and cares a lot about the cost associated with FinOps and, and cloud.

So getting somebody like that to be like, Hey, engineers, Hey leaders. Here's something that it is actually a [00:14:00] focus of ours that we have to actually spend time on, not just the engineering and operational excellence, but also the financial and cost optimization excellence. So having somebody like that who's who you can actually lean on when, when there's things that aren't going your way, or like things that are taking too long, having somebody to lean on like that, that makes all the world of a difference there.

Um, and then the second piece to that is, um, I think. Giving the engineers the tools to, to be able to action these things fully comfortably. Like knowing that all the boxes are checked, and I think that's one of those places why like 0.5 was something that stood out to me was, you're not just getting a recommendation for an EBS on attached volume.

Saying, Hey, this has not been used for 30 days. Delete it. Uh, but you're getting the tools to remediate, you're getting the data behind the, the decision, um, and you're getting the exact script or, or what you need to do to run it. And so I've kind of taken that into my, um, into my, my career and, and my FinOps journey as that's what [00:15:00] I need to get the engineers to do these things.

Not just give them an action or like a Jira ticket, but to give them all the steps that they need. So here's, here's the context behind why we're deleting this. Here's the CloudWatch metrics, or whatever metrics we have showing that it's at 0% utilization. Here's exactly how you remediate it, and here's an engineering leader buy-in, saying he's good with this.

So having all your boxes checked before you take it to them gives them really no reason to say no to it. And maybe they have extra context on why we shouldn't do it, which is fine. But like, if you have all your boxes checked, then You know, You know, you've done your work and your due di due diligence to make sure that these things are getting taken care of and you're, you're being as comprehensive as possible.

Taylor Houck: I, I, I think for the listeners, this is such an important point to really understand, especially if you are a FinOps practitioner, because I think so many people in in FinOps seen as people who just chuck recommendations over [00:16:00] the fence without any context or understanding as to the technical implementations of taking these actions. That's where, as a FinOps person, you need to. Really dig deep into the implications of every recommendation that you're making and when you show up with. recommendation or with a finding, having done that legwork, so you're showing up and adding value in putting things on the table for whoever the engineering resources that you're working with, whether an architect or an engineering leader or You know, an engineer themselves, for them to then consume that content that you brought together for them, almost like you went and did a research project and you're coming to them with all of the things that you found, and then they can apply. All of the context they have in their head from building the application that you're never gonna have, and apply that to all the work that you've done leading up to that point. Don't make them do the legwork. You do the legwork, show up, be collaborative, [00:17:00] and then that's where you're gonna actually see the strong result.

Chandan Bukkapatnam: Exactly, and yeah, I mean, I think it's so important to meet. The, the people that you're working with, where they're at. So like not just giving the engineers like, Hey, this is an action that you have to take, but giving them the cost implications of keeping it on for the next 30 days, next 60 days, whatever it is.

Um, and giving them all the information they need to solve for this. Um, and I think it also comes into like the knowledge piece of it, like. Knowing how to remediate these things, not just from like a pulling a compute optimizer or pulling a recommendation and giving it to the engineers, but knowing exactly how they should go ahead and take care of this.

I think that takes out a really big piece of work for them, and it makes it a lot easier of a decision for them to go ahead and take these things because, yeah, I mean, they can look at your script and be like, yeah, this looks good to me, and they can go ahead and solve for it right away. You, they don't have to do.

Um, a sprint of work for it. They can just kind of take a look at it and be like, okay, that makes sense. Um, and from the finance side, [00:18:00] same thing when you're looking to buy savings plans and things. Maybe the finance people don't necessarily know what a savings plan is, and you have to meet them where they're at and explain these things.

Um, like it was when you first learned about it, like what is a savings plan? What is a reserved instance, and why is it beneficial for us to put upfront dollars from like our budget towards these items to then save money down the road? So I think the biggest thing there is meeting people where they're at.

Taylor Houck: Yeah, you need to be, you need to be adding value to every conversation you have across the organization and bringing. Bringing all of the, the context and the considerations that the quote, like other side of the fence is thinking about. Like when you're talking with finance, you need to be advocating for the engineers.

When you're meeting with engineers, you need to be advocating for finance, but in the language of who you're speaking with and understanding that both sides have a perspective that is important to consider.

Chandan Bukkapatnam: Exactly. Yeah. And I think for me, the biggest thing was like [00:19:00] not being a part of either of these groups before I came into FinOps. Um, it gives me context into both sides and, and like I see how both of them think. And I understand how both of them think. So, uh, I mean like if I had come from software engineering, I understand how I might have a little bit more bias towards engineers or finance vice versa.

So having that like zero experience, zero background in both of these fields, I think it's honestly like in the long run, helped me like kind of spread my wings a little bit and understand both sides equally.

Taylor Houck: It's so important now. Now just switching, switching topics a little bit. I, I'm curious, right? You, you went, you built a, a career, you got a lot of experience working at Hertz. You recently took, um, a, a, a new role, right? I'm curious when you, when you started your new role, um. What were the first things that you were, you were focused on, and if you were giving advice to someone else just getting started, either in their FinOps journey as an individual or at a new company, how would you think about that and, and what would you recommend that someone think about as they go through [00:20:00] that?

Chandan Bukkapatnam: Yeah, I think like, um, for me, when I was switching out of Hertz, I had already had some decent experience in this field, so I wasn't coming into this as a newcomer like I did previously. So my initial steps was I need to get an entire context of the current stack, current infrastructure, exactly what's being used.

Um, and the really funny thing was like. When I, when I came into Procore, I was hoping to find a bunch of optimization opportunities, but I looked at it and I was like, wait, like these guys are pretty optimized. Like, why do they need a FinOps person and like. Slowly, like while I was looking through all of this stuff, you kind of look underneath the sheets where like it's not just the cost optimization recommendations, it's the stuff that makes and drives the decisions.

Like, for example, we were talking about Cloud Trail as one of those things. Um, one of those pieces was cloud trail cost growing linearly. And to me, like being somebody who's been in this field for long enough, I was like, this doesn't make sense to me. Like, I don't see why this needs to grow linearly month over month, like outpacing every other service.

So. [00:21:00] Kind of taking that back to the engineers and, and making sure that I was understanding the entire context, allowed me to then come up with a game plan and give the engineers, here's a step-by-step plan on what you need to do to fig, like, to not just like remediate this, but make sure it doesn't happen again in the future.

Taylor Houck: No, I, I, I want to be mindful. I, I don't want to ask you to, to, You know, speak on any specific company, You know, um, uh, asset or, or divulge any, any private information. But I, I do wanna dig. Little bit. What was it that it was happening with the cloud trail causing it to grow that fast? And were you able to solve it?

Chandan Bukkapatnam: Yeah. Yeah. And I think, I mean, um, this isn't necessarily like private information. I think like this happens across the board when companies like are just kind of being, are trying to be as safe as they can, which is very good practice. Um, but essentially we had, uh, we had CloudTrail buckets or, or CloudTrail going to get all of the requests from all the different buckets and all the different pieces of it.

So one of the things was, um, a Databricks bucket. So obviously with [00:22:00] e femoral workloads coming in and out, um. You have a bunch of requests coming in and out of AWS to Databricks and vice versa. So, and that's not something that you necessarily need because it's just automated stuff that's gonna happen regardless.

And it's safe because it's going from system to system, not from person to person. So like, we didn't need that necessarily, but it wasn't something that somebody went in and did purposefully. They just did it as a holistic piece of like, we need to check for everything for compliance sake. But when you.

Dive a little bit deeper and not just in like the recommendations, but like into the actual services into what they're, what they're going towards. You kind of come across a little bit more of like the stuff like behind the architecture and the configurations. So that was where I found, we found that, um, it was DLH or like data lake buckets that were just sending way too many requests.

So we just use advanced event selectors. Um, so just. Omitting some of these requests that we don't necessarily need to see on a daily basis. So once we did that, we saw 70 [00:23:00] to 80% log reduction. So, um, solve for it pretty much. Like two days in just because it was such a high impact. So, um, like, like I said, quantifying the impact to the engineers.

That was the biggest piece there telling them like, Hey, this is a lot like this is, this is something that like, would allow you to do X, Y, z if we don't have this in place today. Like, you can go self-serve whatever other stuff you want if you solve for this today. And giving them that, that context that like finance would want to give them, but doesn't necessarily know the language in which to give them.

Taylor Houck: This is such an important point because I mean, exactly as you mentioned, this problem would have never, ever. Shown up on Trusted Advisor on any native recommendation engines for cost savings. And unless you were looking for it or were willing to dive into specific services and look at the utilization, ask questions, where is this coming from?

Why is it set up in this way? You would've never found this optimization [00:24:00] opportunity, which ended up being, uh, it sounds to be. impact on, on the cost of, of this specific application or, uh, or workload.

Chandan Bukkapatnam: Yeah, a hundred percent. Um, and I think like when somebody's starting out at a new company in FinOps, like coming back to that question, um, I think this is one of those things that really helps build trust. With you and the engineering folks that like, Hey, I know what I'm doing. Like I, I'm not just giving you guys these recommendations out of an engine.

Like I'm actually looking at our usage, our optimization, um, opportunities beyond just trusted advisor and these recommendations. So giving them the, the sanity that, or the sanity check that I'm ticking all my boxes. Like you can go look at exactly what I looked at, follow my workflow, follow what I just did, and I give you a document on, and you can get to the exact result that I did.

And, and kind of giving them that context of here's how you find what I found and here's how you remediate it. And like giving them all of that information makes it a really, um, [00:25:00] hard thing to say no to as well. Like that's, I mean, and that's, that's our job is, is to give them these recommendations without having the, um, the leadership or, or the ability to be like, you have to do this.

Like, you just have to be able to give them all the information necessary without having that authority.

Taylor Houck: Yeah. The name of the game is influenced without authority.

Chandan Bukkapatnam: Yep. Mm-hmm.

Taylor Houck: that we have, like the feather in our calf is that we're talking about dollars and cents.

Chandan Bukkapatnam: Yes.

Taylor Houck: One of the things that, like engineers, they, they very frequently don't have, um. A, a demonstrable, quantitative financial impact on, on a

Chandan Bukkapatnam: Mm-hmm.

Taylor Houck: It's very rare that they're able to say, I drove x in, in, in value, right? Based on the, the, the feature that I shipped, or based on, You know, this application that I developed. And I think that everyone knows the work they do is really valuable, but being able to quantify something like a cost optimization, a cost optimization initiative, and show the big impact that you made, and then like. I'm curious what, how, how [00:26:00] you deal with this, but like for me, in my experience, I always like to really them out and, and give them the credit for, for taking these actions and, and showing the value that, that they drove through these different projects.

Chandan Bukkapatnam: Yeah, no, I, I completely agree. Um, I, so we, what we do now is we kind of have like a weekly cost, cost pacer kind of conversation. Um, and within that, like, You know, when we're going through our weekly updates, I always like to, to just go through a list of the cost optimization, uh, opportunities for that week.

Like stuff that we found, stuff that's materialized, um, and giving them. The total shout out, um, of like, without these guys taking the initiative to just take some time out of their day to do this, we wouldn't have done X, Y, Z and savings or X, Y, Z and like cost avoidance for the month. So giving them that in front of like a panel of like leadership that gives them more motivation to then go and actually.

Seek you out to go find more stuff to optimize within their domains and their, and their like, [00:27:00] subject matter expertise. Um, and I've seen that firsthand where they'll come to me and they'll be like, Hey, can you go take a look at this database? Like, I don't think it's being optimized. I think it's underutilized.

Then I can go and be like, oh yeah, yeah, that you're right. Like that's, that's correct. And. That's when I think you start to see a lot more success is where you're not the only one finding the recommendations. Everybody's finding the recommendations and whether or not they surface them to you is, is non-material as, as long as they're going ahead and taking action and bringing those up to their leadership.

And I, I think that's where you start to see more of, like in the crawl, walk, run, you start to see a lot of the walk in the run there.

Taylor Houck: Yeah, it's getting it. It's really building that cost aware culture, and this is where. Um, it's interesting because you talk about like the, the kind of like the hard skills and then the soft skills of FinOps. Right? And like, in my opinion, you, you really need both. In this conversation we kind of started with, with the hard skills.

To me that is a prerequisite. Like you must be able to show up with a, with a, an informed opinion, show up with the data points, understand the implications of what you're [00:28:00] saying. Like if you don't have that, it's a non-starter. You'll never be able to build the relationships that. You know, that, that, that enable the soft skills that the culture side to really emerge.

Chandan Bukkapatnam: Exactly.

Taylor Houck: the combination of the two is where things get really powerful.

Chandan Bukkapatnam: Yeah, and I think, I mean, that comes back to what we talked about earlier, like meeting them where they're at. Like we don't, we don't necessarily need to to sit here and explain to finance why we're using M four eight larges, or X two I idea and 16 x larges. You just have to tell them the business impact of not using these or using them and, and giving them some KPIs that they can then take back to their leadership and be like, Hey, we're using this much of this because our, uh, our, this is a, this is what our percentage of our cost per revenue is, and all these different KPIs that they'd be interested in.

Um, and giving them that context there. Makes all the difference when you're doing your work so that they can also just kind of give you the support you need, um, in making decisions like savings plans and all of these things, giving them what our ESR is, our [00:29:00] effective savings rate, um, and what that means, what that differentiates us from other companies by.

And just kind of giving them much more context than they initially had. I think that's the biggest change, um, that I think FinOps has, has given most companies that I've, I've seen worked with.

Taylor Houck: Awesome stuff, man. Awesome stuff. Now I'm gonna switch gears one more time and I wanna, wanna put you on the, on the spot a little bit about a topic that everyone's talking about these days, and that topic is ai.

Chandan Bukkapatnam: Yep.

Taylor Houck: I'm not sure if you saw, and, and by the time this, this recording, this podcast comes out, it's gonna be a little bit, a little bit dated, but just this week at the time of our recording, we're recording on. Friday, February 20th, the 2026 State of FinOps data came out and it showed that a lot of organizations see ai, both FinOps for AI and FinOps, uh, AI for FinOps as top priorities heading into 2026. How do you see AI changing, um, this industry?

Chandan Bukkapatnam: Yeah, I mean, like you said, so like FinOps four [00:30:00] ai, ai four FinOps, those are the, the two buckets that we see. So, um, FinOps four ai, we could start there. Um, that's, that's one of those things as companies start to implement a lot more, um, AI into their workflows, you, you don't really see that quantified. Um, you see a lot of these.

Um, these new workloads, these new like, like, uh, open search and all of these different services, bedrock coming in. Um, and you can't really put a KPI, you can't put a face to the name on those. Like, what is this bringing us, like, like value wise? Um, and that's, it's kind of like when you initially start with FinOps at a company, a company that's never heard of FinOps, coming up with KPIs that matter to them.

That's, that's where you start off. So with FinOps for ai, that's kind of where I, I want to, I want to start off is what does this mean for us? Like, what is, what is a cost, cost per engineer? Like, what is, um, the cost per token for us? And, and as you start to implement these things, you start to see a little bit more of like.

Confluence with decisions that are, that are being made on like a, [00:31:00] a top level. So on, on an optimization scale, it's a little bit tougher because companies are kind of okay right now with being over-provisioned. They want to implement more, they want to do more. So it's a little bit more of like a, a place where we're at a standstill where.

We have some of these things that are unoptimized, but it's okay for the time being because we want to focus more on the operational excellence of it rather than the cost excellence. And once we get to a place where the cost or uh, operational excellence is good enough, then we can start to focus in on, on getting these lower levels of optimization done.

And, um, as far as AI for FinOps. That's a piece that I'm really excited for as well. Uh, I mean, like, obviously, You know, being a FinOps practitioner, um, I don't see it ever taking away FinOps roles. I see it, uh, like, like ly amplifying him and, and making your job a little bit easier, um, and not having to focus so much on the nitty gritty of stuff.

Like I don't, when [00:32:00] I, when I think about the favorite parts of FinOps for me, I'm not really thinking about going in and looking at unattached DBS volumes and these kind of stuff. I would rather build out a workflow, not just an automation workflow, but a smarter automation workflow that has these things thinking.

On when to delete these and looking at when they should not delete them and looking and making these decisions without me having to manually go in and make all these little changes and approval flows and, and all of that stuff. So I think that's where FinOps for AI starts off really well. Um, is, or AI for FinOps, sorry, starts off really well, is getting all those nitty gritty stuff out of our, our peripheral.

So just making sure that those are all taken care of and then you start to go into a little bit deeper.

Taylor Houck: Yeah, I think it's gonna become a much more strategic function and

Chandan Bukkapatnam: Yeah.

Taylor Houck: about this before and it's like interesting because both. Sides of the FinOps for AI and AI

Chandan Bukkapatnam: Mm-hmm.

Taylor Houck: They're both, they're, they're distinct and separate, but they're also very related, [00:33:00] right? Because when you think about FinOps for ai, right?

I am managing, I am optimizing AI workloads. Well, the spend on these workloads is only going to be increasing probably at a very rapid rate,

Chandan Bukkapatnam: Mm-hmm.

Taylor Houck: what that means is that the amount of spend. That we're, we're putting towards compute or cloud, or however you want to think about it, is going to be increasing likely, uh, rapidly.

Right. So that increases the scope of the role as you're managing more spend. Right. And also it's, it's new pricing models. It's, it's new, uh, new services, new platforms. Right. I mean, you're seeing. I can imagine there are companies out there that are spending more on Databricks and Snowflake and Anthropic and OpenAI than they even are on AWS, especially the newer companies.

And you're gonna see this trend more and more. So it's like understanding these aspects and being able to manage and optimize that at a, at a high level is very important. But at the same time, many of the tasks and the functions that have traditional. person is doing is, is probably gonna [00:34:00] get disrupted with AI

Chandan Bukkapatnam: Mm-hmm.

Taylor Houck: Right? So then where, where is the human, You know, adding value that is, that is unique and distinct. And it's gonna be a, a really interesting trend to watch. I've got a lot of

Chandan Bukkapatnam: Yeah.

Taylor Houck: Um, but it's, but seeing it all play out is gonna be where it all, it all comes together.

Chandan Bukkapatnam: A hundred percent. And I mean, yeah, like obviously AI is, is taking over the world and everything, but context is still number one. And, and in FinOps that's kind of the piece that we stand out in where we're providing context to, to why we can or can't delete certain things. As well as the engineers, they're providing us context with why we can or can't delete or optimize certain workloads.

Maybe they're over-provisioned for a reason, maybe they're not. Maybe they're just autoscaling groups that AWS or the AI engine thinks that they're just over-provisioned. 'cause they ran, they came in and they came back down. Like this is the kind of context that we're still not there yet where they can just do all of that information building and gathering and taking the decisions.

So I think, I mean, as we move on towards a little bit [00:35:00] more of like the A GI world, maybe, maybe we see that. But I am, by then, I mean, who knows?

Taylor Houck: Well, Chandan, I'm, I'm super excited to see how it all plays out. And, and one thing is for sure, I know, I know you're gonna be there,

Chandan Bukkapatnam: Oh yeah.

Taylor Houck: To be a part of it and, and pushing it forward. So, I mean, perhaps we'll jump on another one of these in a year and we, we can look back and, and laugh at, at what we said, or maybe, uh, we'll have some, some new thoughts or insights.

But one, one thing that I, I really believe is, is AI is gonna be a big. Multiplier and make, make people who really embrace it a lot more, impactful, right? And, and, and let you pretty much get a huge, You know, way, way more output and productivity out of someone who, who's willing to push on it. So, um, I'm excited to see how, how we all embrace it and, and, and how it all plays out.

But, but AI will, will never, it, it's not human.

Chandan Bukkapatnam: Yeah.

Taylor Houck: And, and, and we're gonna be, You know, a part of this no matter what. Um, but kind of, uh, pivoting, pivoting on that. You know, there, there, there, there are certain [00:36:00] things that are intrinsically

human,

Chandan Bukkapatnam: Mm-hmm.

Taylor Houck: And one of those is sport, right? I know that you're a, you're a big sports

guy.

Chandan Bukkapatnam: Yeah.

Taylor Houck: Play sports. You watch sports. What, what, what are your, uh, teams and, and where do you like to, to play?

Chandan Bukkapatnam: Yeah, no, a little bit of both for sure. But, uh, basketball and tennis are my vices. Um, love watching basketball. Uh, love watching tennis. Uh, and I, yeah, we talked about this, but basketball obviously, You know, being from Denver kind of have to support the nuggets. Um, and it's, it's not hard to support the nuggets either.

Um, I was born in Portland, so I supported Trailblazers for a while. Um, and that was a tough one. I mean, other than, other than Dame, it was really hard to just, You know, support them. Uh, but the nuggets are easier. Um, and, and You know, with tennis, like we've seen some, some great matches come out of the newer.

Newer folks and see, saw some of the older, um, most accomplished athletes kind of retire and it was sad to see, but no, it's still still amazing to see.

Taylor Houck: Tennis is interesting, right? Because it's like, I feel like you've got these, these names and these people that have been [00:37:00] around for seemingly forever in the

sport, right? Um, like Na Doll Federer. But then now you've got, You know, some new kids coming on the block that are taking care of business,

right?

Chandan Bukkapatnam: Oh yeah, yeah. No, a hundred percent. I mean, and, and you see some of these guys like Alcaraz, uh, Ben Shelton, all these guys, crazy athletic. Like I've never seen people that athletic before, like the way that they just, they'll like run from side to side and then back and it's just, it's insane to see like the level of athleticism growth throughout all of sports throughout the years.

And especially in tennis, you see that a lot where. They, they kind of just have to like, continuously just sprint across the, the court now. It's amazing to see.

Taylor Houck: Yeah, it's interesting how it's like in pretty much every aspect, it's like the generations just keep getting. Better. Right? It's like, and I, I'm wondering if that's ever gonna slow down or stop, but it's almost, it, it, it's almost like, not to be too cliche or corny, but it like ties into the professional aspects

as Mm-hmm.

Where people, uh, as [00:38:00] individuals are gonna be doing more and, and, and. More productive than there were in the past. Now in our sense, we're, we're enabled by technology in the athlete sense. I mean, they are kind of too, I mean

you

Chandan Bukkapatnam: Yeah.

Taylor Houck: the Tom Brady method, like

what

Chandan Bukkapatnam: Yeah,

Taylor Houck: with this body. It's like these guys in the 1980s were not doing this.

They

Chandan Bukkapatnam: exactly. Yeah.

Taylor Houck: Just eating whatever they wanted. They were not really working out. And now these guys, um, are like Training their like 24 7.

Chandan Bukkapatnam: Yeah.

Taylor Houck: Their body like a, like how much LeBron's like investing, how much per year in Millions. Millions. Yeah.

Chandan Bukkapatnam: No, it's, yeah, it's, it's crazy. And I think like, even on like the most like mundane level, just even being able to access so many different trainers and different people, like, didn't have that back in the day now, like, these guys are trained by the best of the best from a young age.

So it's, it's really cool to see like the evolution of sports too. And like tennis used to be with a, with a racket, like about the size of your, like your hands. And now it's huge. And [00:39:00] now it's, it's really cool to see.

Taylor Houck: Awesome, awesome stuff, man. Well, well thank you so much for joining the show. I really enjoyed having you on.

Chandan Bukkapatnam: Yeah. No, me as well, man. And, and thanks for having me.

Taylor Houck: Yeah. And to the listeners, I hope that you enjoyed another amazing episode of, of FinOps in Action, uh, really insightful episode, Chandan, thank you so much for, for coming

on.

Chandan Bukkapatnam: Of course.

Taylor Houck: Until 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, empowering teams to optimize cloud costs with deep detection remediation tools that actually drive action.