FinOps in Action

FIA - Chitra Jadhav
===

Intro: [00:00:00] 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 has spent the last eight years building FinOps programs from the ground up. Along the way, she has delivered real results, building cost transparency down to the product level, increasing tagging compliance from less than twenty percent to over ninety percent, and driving roughly nine million dollars in annual savings.

She is a great example of a multidisciplinary FinOps leader with an engineering background and an MBA who has applied her expertise both as an [00:01:00] operator and as a consultant across multiple organizations. Today, she is a senior manager at Rogers. Welcome to the show, Chitra Jadhav.

Chitra Jadhav: Thank you, Taylor. Thank you for having me and giving me this opportunity. Looking forward to the discussion.

Taylor Houck: absolutely, Chitra. I'm super excited to be chatting with you today, especially because of just the pure number of organizations in which you've implemented FinOps practices within. When you look back at all the organizations that you've worked with, what's the number one thing that you wish new organizations understood about FinOps?

Chitra Jadhav: So, um, I feel that the number one thing that all of the organizations need to know that in order to achieve big in FinOps, you need to make sure that at the get-go, there is a recognition for the need to have a FinOps body in the first place, having that requirement, uh, sorted out in your head and also knowing that you need to have a strong governance, meaning how do you [00:02:00] set it all up to help FinOps self-sustain.

Taylor Houck: If you were going to describe to a leader of an engineering organization why they need FinOps, how would you describe that to them?

Chitra Jadhav: So on the cloud, it gets, um, very, uh, complex when we are provisioning, uh, resources rapidly and, um, there, there's a lot of, uh, infrastructure that is multiplied as we speak. There are deployments happening and without having a FinOps body that can help you keep track of is traceable to where and what resources could be attribute to-- attributed to which cost center becomes extremely key because otherwise you'll just receive a huge invoice and will back off from paying the bill because now you don't know whether you are justified in paying that big chunk.

So engineering team needs to understand that, um, everything has to be properly broken down, um, and that invoice has to make sense, the run rate has to [00:03:00] make sense, and in order to do that, we need to make sure that FinOps is at the table.

Taylor Houck: Such a good point. And, and one other word that you touched on, uh, before was governance. But before we get to that, I wanna talk about starting FinOps practices from the ground up, because I understand that you have done this multiple times now. When you walk into an organization that has never had a FinOps program before, what's the first thing that you do?

Chitra Jadhav: I generally begin with trying to assess what the current state gaps are, and that usually entails using the awesome material that FinOps Foundation has also given us. Uh, it's about capturing the maturity assessment, the maturity score of where the organization is at various stages of, you know, FinOps pillars, like where are you with your inform capabilities, your operate capabilities, and your monitor capabilities.

And then using that score, you can benchmark against, you know, how you can position yourself, whether you're at a walk stage or a crawl stage, or you're ready to run or even in a fly [00:04:00] phase, depending on how advanced and mature your organization already is. So it helps you, uh, gain that visibility, and that usually is the starting point to kick off any FinOps practice to have a solid foundation.

Taylor Houck: And once you establish that visibility and, and put it in place, what, what comes next?

Chitra Jadhav: it usually entails about how is FinOps going to in- interact with the rest of the key stakeholders, because FinOps on its own can do very little. We have to make sure that we have ongoing and frequent, uh, engagements with your finance team, with your engineering team, the provisioning team, then also, you know, the application teams, sometimes even the procurement team, because, you know, for doing your cloud, uh, synergies or tool synergies, sometimes you could get a good deal by working closely with them.

So that's how slowly the entire ecosystem evolves to, to show that FinOps is, you know, going to fit into the entire model and everybody [00:05:00] has a role to play in, in FinOps. It's not just the FinOps consultant or the FinOps leader who would be in charge of driving the FinOps objectives, but every stakeholder involved in the cloud ecosystem has a role to play.

because of that, there are... There's so much job to be done from so many perspectives by so many players, the role players.

Taylor Houck: No. It really is a, um, a, a team sport, not an individual sport. But in your experience, who, who is FinOps rolled up to? Like, is it an engineering function? Is it a finance function? Can it succeed in both? Do you have a preferred, uh, method?

Chitra Jadhav: Well, I would say on its own FinOps should be declared as its own body. But that said, it definitely touches all aspects, like I just said. So with finance, I think finance is mainly focusing on the CapEx aspects of it, wherein you're looking at your third-party labor costs, the licensing costs, and maybe the internal and [00:06:00] external, uh, services that you're availing.

But that's again a very, uh, I guess, uh, restricted view of where your cost is coming. But in order to boil down how the metrics actually matter from a bottom-line perspective and from a top-line perspective, you have to do much more than that. And those metrics are gathered by working in tandem with your development teams, your engineering teams, um, so that we can, you know, uh, plan the performance of, of a balanced workload.

Your workload is not just about, um, you know, your speed of the delivery, it's also got to do with whether it's meeting the governance standards or not. So it all kind of has a holistic impact, and that is why we have to make sure that all of the stakeholders are doing their part.

Taylor Houck: And there we go, back on that governance topic again, and I know this is a topic that is near and dear to your heart. I think that that word,

Chitra Jadhav: Yeah.

Taylor Houck: governance, especially [00:07:00] from FinOps personas, can mean many different things. Can you define what you mean when you say governance?

Chitra Jadhav: So when I say governance, what I mean is, is there a standard established and a repeatable process that you could do over and over again, which is generating the same results and the o- the outcome that you want? Meaning, like, um, uh, the various stakeholders that you're engaging with at various stages of your, uh, delivery.

It could be maybe if you're an IT setup, you could talk about the SDLC, like every, every part of your SDLC, whether it's the design stage, if you're dealing with an architect or a designer or a developer or a QA person, or from a more agile standpoint, you're working with a product owner. Everybody have their part to do, and some of the audit or checkpoints that FinOps does, uh, to integrate with, uh, these stakeholders is something like having a seat at the ARB table, wherein every intake is assessed to see [00:08:00] whether the submitter of that particular initiative or intake has considered the costs in mind before, you know, trying to get approval for that particular intake.

Because we don't want to, you know, have something, uh, to be built that may end up costing you thousands and thousands of dollars if that could have been done at a, you know, more optim- optimal way. So that is why it becomes key that you have these checkpoints where FinOps engages at multiple points. ARB is just one that I mentioned, and then there's also the PRR stage Which is called as production readiness reviews, wherein again, just before you do your final deployment, you are also making one final sanity check to ensure that your tagging is as per, you know, the appropriate standards and the nomenclature.

It's following certain compliance, uh, initiatives that your organization wants you to do, and also from the perspective of, uh, tracing back your resources and how [00:09:00] attributal, attributable those costs are. So all of those checkpoints, uh, are kind of, uh, put in by FinOps. And also the most important key is the provisioning, right?

So provisioning is where, um, things could, become, uh, uh, complex very, very quickly, uh, because people do provisioning multiple ways. We do it through automated way, or we could do it through a self-managed way. And that kind of dictates how your, uh, you know, costs are gonna look like eventually because then you, you realize that if you've not done your tagging properly or if, if you've not put in some key FinOps checkpoints in place, you may be in a situation where your chargeback is suddenly failing or maybe your for-forecast is not as expected or maybe an optimization has become, uh, a guesswork for you now because you've not or, or your engineers have not followed a certain protocol.

Taylor Houck: Getting involved in the actual deployment of resources. Th-this must be how [00:10:00] you, uh, went about bringing your tagging coverage from twenty to, to eighty percent. But before moving on, you, you mentioned ARB. I believe what you meant by ARB is the Architecture Review Board. Is that right?

Chitra Jadhav: Yes.

Taylor Houck: Can you describe what an architecture review board is and why it's important?

Chitra Jadhav: So ARB is not just Oriented, uh, to FinOps, but it kind of addresses many parameters associated with an intake. There are many stakeholders from various departments who kind of, uh, join this call, and it's usually run by, um, coordinator who's, uh, uh, who ask these folks to come up with two or three key trigger questions that are key for their, um, intake to succeed from their particular department perspective.

So, for example, if there is a cyber person who comes in, he's making sure that the two or three questions related to cyber that are absolutely key are answered, and ditto for [00:11:00] FinOps. So it is something that I have done from my perspective in my company called Dayforce, where I j-- was working before my time at Rogers.

So, uh, during, uh, these discussions, usually we would, uh, be able to determine if our intake was a new build or whether it was an enhancement to a existing initiative and whether a proper cost estimation template was, you know, submitted by the product owner, whether they even, you know, like, took a note that they should be using a cost calculator, for instance, which is readily available on our, um, provider websites to get at least a near close, uh, approximation of where their overall build is going to end up.

situations when some engineers wouldn't even know the holistic size of what they are trying to build, even though from their perspective, they say, "Oh, we just need three VMs and one storage." But when you put everything two and two together to, you know, just launch that and build it, it's, it's not just, you know, s- [00:12:00] dollars and cents we are talking about.

It's huge. And, um, and knowing that is key because you're keeping your developer aware, cost aware. And cost awareness is a kind of culture that is pivotal for FinOps to take off or to take flight successfully.

Taylor Houck: This is actually a really interesting concept and one that I have not heard from many others. I'm curious with this ARB, number one, how frequently are you guys meeting? Two, is it a synchronous chat? Like, are you actually jumping on a call or getting in a meeting room and, and meeting with these folks?

And then based on that, how do you make sure that you are being, let's say, value additive and, and, and not just adding additional process and bloat? Because I know that if every time an engineer wants to deploy a new resource, they have to sit in an ARB meeting, that, that could slow things down, but there's obvious benefits to doing it.

So I, I'd just love to hear how you make this work.

Chitra Jadhav: [00:13:00] thank you for asking. So ARB, uh, is used usually, at least at Dayforce, the way it used to be conducted was we used to have a series of sessions.

I

The intake comes in in the beginning, and then people are trying to assess what the purpose of that intake is, what are the kind of requirements that, uh, hardware or software requirements it, it's gonna require to succeed, and so on.

So all that is listed out. And then there comes a another round two, in which all of these engaging stakeholders participate and then ask their set of questions. So from a FinOps perspective, when I join in, my trigger questions are gonna be like: Have you put an appropriate, uh, tag meta value, uh, labels as specified by my tagging guideline nomenclature document, which I was...

which I have already published and it's in a certain location on the Confluence. Have you gone there and, you know, done your part wherein you've adhered to those tag value pairs, the mandatory tags or, you know, the tag meta value [00:14:00] data, like whether it's a standard nomenclature. I... We don't want, you know, versions of prod, production or non-prod, non-production floating around because that becomes very difficult when we are trying to filter down or trace down cost to a single source.

So following those key standards are necessary. So that could be one of the questions. The other one could be, have you done your cost estimation of all the components that make up your intake? Have you... Do you know in a ballpark range what would be the init-initiat-- the cost of this initiative? And then, uh, based on that, maybe it can open a conversation with the finance team, whether it's even in their budget to fund it or their specific internal finance team, if they are okay to fund that initiative or not.

So those are some examples of, um, questions. And then, of course, the cyber team on their side, they have different set of questions from a security standpoint and so on. So all of us, we, um, participate, and we make sure that any initiative before it gets approved and people [00:15:00] start building it, uh, it's, it's been, uh, looked at from a holistic standpoint that you don't end up building something that's gonna end up becoming a white elephant for the rest of the company.

So that's our, you know, way of making sure that everybody has, uh, participated.

Taylor Houck: think it's so important, uh, that FinOps has a seat at that table because I can imagine in a lot of, uh, instances, cost is more of an afterthought rather than something that's being considered throughout, um, the planning cycles.

Chitra Jadhav: Very true. Um, it's, it's important that we have a presence there. Otherwise, uh, the work becomes even more difficult to isolate when we are already in a reactive stage where people have done their part and then we are suddenly going back to check, "Oh, why has this workload costed so much? That's like burning money and we really need to do something to see if we can, um, make it more optimal or if there are scope for opportunities to reduce it or not."

So instead of, you know, like, um, [00:16:00] paying those, that money and incurring that waste, you are taking some steps proactively by making sure that you are present at these discussions.

Taylor Houck: Absolutely. I, I do want to now kind of shift the conversation to post-deployment and actually talk about optimization because the, the, the shift left narrative, it's so important and as you're mentioning, if you can avoid wasted costs from occurring before it even happens, that's the best case scenario,

Chitra Jadhav: Yep.

Taylor Houck: But due to the flexible nature of the cloud, the way that these services work is there's so many different resource configurations and architectures, and actually the optimal decision, right, the optimal, let's call it resource configuration, is highly dependent on how the resource is utilized, right? The utilization.

And that utilization, one of the benefits of the cloud is that it inherently changes over time and you can adjust to that.

Chitra Jadhav: Yes.

Taylor Houck: I want to talk about identifying and executing [00:17:00] optimization opportunities. In your experience, what is the, the best way for FinOps to be involved and work with engineers to execute these initiatives in a way that is, uh, being a, a good steward of the, the company's capital, but also respecting the, uh, the, the risk that is sometimes inherent in making changes to your, uh, cloud resources?

Chitra Jadhav: that's the exact idea of FinOps. We want things to be scalable, uh, as per the needs also. We don't want to say that, "Okay, this is the

There's multiple

case. You can always underprovision or you can, uh, scale up or scale down depending on the growing needs or the changing needs of, uh, your, uh, initiative.

um What works best is having a FinOps champion model wherein if especially you're working with like [00:18:00] one fifty developer teams, at least you have a champion representing each of those teams who could be working with you in maybe office hour kind of a setup wherein they can join in any time and let you know that, "Hey, by the way, I need help and see if we can, you know, reduce the, uh, the wastage on this particular subscription or workload because the utilization seems to have dropped, um, from a certain percentage to certain percentage."

And by the way, we could track that, right? Like, uh, the, the, uh, assumption is that FinOps has already done everything out there to ensure that we are tracking your real-time consumption using our, um, like your, your Power BI reports are... I guess they, they do the ballpark of all of your work. Even if you have something as basic as Power BI, you're already ensuring that every spend that's happening on the cloud, you're tracking it live through reports.

You're able to, you know, go to a certain granularity back to the resource level, resource group level, and [00:19:00] so on, and you, you can see exactly how much your resources are consuming. And, um, having, having all that things already in place, you are in a position to work with that FinOps champion to determine if they need to make some changes to how their s-subscription has been set up, if there are some VMs probably that they should be decommissioning because, uh, they've been sitting idle probably from a certain time.

So there are many combinations of optimizations that could be done. Uh, some, um, could be through your automated scripts using Lambda functions where you could, you know, detect idle wastage, maybe turn off VMs or, uh, resources that are running wastefully during holidays or after hours turn them back on.

It's more like a serverless provisioning which can help you, you know, like, uh, be lean on the cloud. Or it could be h- you know, setting up alert notifications where you know that a certain threshold is acceptable, but [00:20:00] beyond which there's certainly something that's going wrong, so you have to dig deeper as to why that resource is costing so and so on a certain date.

So enabling all these mechanisms, uh, in combination with having something like a automation script in place, it, it kind of puts every... It's kind of, you know, helps you to optimize a lot of dollars that way. Um, so you have to do your part. There's not one, uh, method for, for optimization. There are multiple things that you could be doing.

Taylor Houck: Multiple methods for optimization. There's also multiple ways in which organizations handle this. I want to go back to just the, the pure fact that you have done FinOps at so many different companies. I mean, do you know how many you've, you've done FinOps for, all things considered?

Chitra Jadhav: at least six plus now. Yeah.

Taylor Houck: plus. So I'm sure you've seen different programs and organizations where FinOps has more authority or less authority, and times where FinOps maybe ha- i- is more [00:21:00] risk-averse or more risk on.

How have you seen this play out and, and do you have a perspective on where the, the optimal position is?

Chitra Jadhav: That's a great question because the reality of it is that FinOps is as powerful as how much power you're willing to give it, right? Like, we could be s- sidelined if you treat us that way, and maybe we could be like a dying voice there, and s- that won't be effective at all. Or you could keep us at the table with the rest of the stakeholders and make sure that you're integrating our decisions in your day-to-day.

You're helping us build a culture of a FinOps mindset. There are some companies in which I have seen, uh, personally, my leaders who've been, um, you know, given, uh, good p- lot of permissions and rights to the FinOps team to go ahead and maybe kill a resource if they feel that it's been going havoc on the cloud.

There's no accountability to it despite, you know, um, [00:22:00] warnings given to the product owner and so on. So we had that autonomy. Uh, as a FinOps consultant, I did that at Thales when I worked there. So it depends o- completely on how bold your leader is and how much trust and faith they keep in FinOps to let them carry on their task.

And at the end of the day, I feel that even FinOps has to buy that trust, right? Like, when we have established the solid governance wherein we are showing you that, by the way, these are the protocols that we follow." It's not just that I'm just gonna go into a certain account and delete your resources on a random day or in production environment.

That's not gonna happen. So even we have our own prot- protocol, our rules. So buying that trust and making everybody aware that we are here to help. We are not here to, you know, um, pose a problem to your business as usual activities. Instead, we are here to ensure that your team looks good from, um, you know, an e- team economic point of view or from a budget consumption point of view as well.

So that kind of mindset has [00:23:00] to be adopted. And then I have worked in companies wherein FinOps was kept, um, not at the forefront. It was more to do with like... It became like, uh, are we the ones who are trying to put people in the line to stay within the line. So, you know, having that kind of a approach to FinOps doesn't take you far because, um- we, we are always in a position where we have to justify, why they need to, you know, cut down a certain spend or why are-- why they should be considering cer-certain set of recommendations.

But you will make our task much more easier by being on board with, uh, what FinOps is all about, and that we are not here to make you look bad, that it's not a wall of shame that we are trying to create by having all those dashboards that are revealing the numbers. But we are here to make sure that everything is done in a value-oriented fashion, that there is minimal wastage.

Even though we are not saving you thousands and thousands of dollars, we are at [00:24:00] least making sure that for every dollar that you're spending, you get the maximum ROI from it.

Taylor Houck: And we need to be a partner to every single group within the organization, and everyone should be feeling like FinOps is value additive, right? And, and one of the things that I see FinOps teams struggle with is when they are perceived as You know, being a thorn in the side or playing a blame game rather than being, you know, a trusted partner.

And I think that's where ensuring that your FinOps team and your FinOps program is full of people with a collaborative mindset and also people that have deep, real expertise on the engineering side is also very important because that's how you build credibility and that's how you can be successful in winning the hearts and the minds of the folks that you work with.

Chitra Jadhav: I completely second [00:25:00] that. Uh, also there are, uh, many misnomers that are attributed to FinOps. Like for example, at least with the tagging that I worked on, we were expected to have hundred percent compliance. Now, that's very idealistic, but, uh, in real time, if you are working in a co-uh, co-organization which has a combination of automation provisioning as well as self-managed, that's like, uh, you need to wave goodbye to that already because the moment you have self-managed infrastructure, uh, it actually translates to create anything, anywhere, anytime without tagging, and that itself invites many entry points for non-compliance.

So you have to be realistic. when you declare that, uh, big goal, always have a reality check as to what's achievable versus what's an undue pressure that you're trying to put on the FinOps. Likewise, uh, if I were to even put the tagging conversation aside also from the perspective of cost savings, right?

Like blindly before we [00:26:00] build a roadmap, we are asked to give a percentage of how many percentage savings we are gonna give you after this annual, um, you know, year ends. So while we can give that an approximation, what we also need to realize is that here we are trying to Manage costs that are already sunk, meaning you've already signed that big budget dollar with your CSP wherein you have to adhere to that commit dollars over the next four or five years.

And if you go under the commit or over the commit, it's both going to get you into a penalty, right? Because you have to stay in line and not overspend or underspend because anyway it's a sunk cost and you will end up paying the penalty regardless. So it's about balancing the optimizations and balancing your, your savings based on how you're trending, how your consumption run rate is trending.

And you can always play it by the year, right? Your strategy has to be com-- [00:27:00] always evolving, always recalibrating to, you know, like, uh, take size of your current state.

Taylor Houck: Yeah.

There's a lot of different scenarios that you can find yourselves in where priorities can change over time. Um, and, and optimization is not always the top priority for a FinOps team, but in my experience and from chatting with people, it oftentimes is, right? A lot of the times what we are doing is trying to, uh, let's say either reduce our, our total cost, our run rate, or even make room in the budget for new programs that we wanna build out, right?

I mean, that's what at NBC Universal we would call it self-funding growth because we would always have a lot of new applications or new workloads that we wanted to bring online, but we didn't have the budget to do that, right? So rather than having to go back to the finance team and advocate for additional budget to build out, you know, these, these applications that the product teams, the business teams were demanding, we would try and find [00:28:00] that budget through optimizing our existing infrastructure.

Now You know, you've been doing FinOps for a while, that there's a lot of recommendations that will come out of the cloud providers themselves that tell you how to save money, right? And, and oftentimes there's a lot of good insights to be gleaned from that, right?

Chitra Jadhav: Yeah.

Taylor Houck: my experience though, uh, it's, it's takes a relatively short amount of time to work through those, right?

Chitra Jadhav: Yeah.

Taylor Houck: and once you've gotten the value out of the, the native cloud provider recommendations, you're left in a position where you have to go and, and find new opportunity for yourself. I'm wondering if you have come across this scenario, uh, through your days in FinOps, and if so, how you, how you managed to find opportunities to optimize.

Chitra Jadhav: So I, I like to start with, um, you know, what we were doing as a FinOps team and then, um, at an organization level. So basically we-- I [00:29:00] was working, uh, internally to a cloud team, but at, um, COE level, there was this one of the Big Five consultants who had to step in. They were actually brought in to see if they could do more in terms of, you know, optimizing our cloud spend.

So then of course, they came up with these, uh, so many recommendations that, you know, "Hey, do this, do that. This is a unnecessary workload. Reduce your, uh, VM size here or, you know, uh, have this particular family type on." But then when we were actually looking at the pragmatic aspect of whether those recommendations actually apply, so when we, you know, did all the screening, we rel-we realized that they don't fit the context, right?

sometimes you need to have... There, there are certain decisions that engineers or their leaders do take because that's how their team is familiar with working or it could even be that, you know, that's the third-party products that they are interacting with and because of synchronicity, they want to go a certain way.

So there are many, many, uh, criteria that go in which [00:30:00] cannot, you know, meet the eye by just a external consultant coming in and, you know, giving you, um, a top advice or maybe even, um, third-party analy-analytics tool, uh, like even the advisor that offers. So those recommendations don't tend to be legit when you actually go down to the depth of it and you see that, no, you absolutely need a certain size of this VM to be there for so and so reason.

So doing that kind of a analysis, um, has always been time-consuming and, uh, what I was hoping with the AI and machine learning coming in as a added benefit to FinOps is if we are able to, you know, have a smarter way training the tool in such a way that you ignore certain types of, uh, specifications of your resources.

They have to be of a certain, you know, specification and dimension. Keep that in mind, and then you provide a recommendation that's more value-based on top of that, right? So that's yet to happen because, uh, I guess the current, [00:31:00] uh, internal tools in all of the CSPs are, uh, not having that, uh, AI or, uh, ML component embedded in it.

And that's why there's a lot of, um, third-party tools, uh, in the market today that are, you know, saying that they can do better, like this NetOps that I've heard of. Cloudability, uh, I guess that needs to do a little bit more to adapt AI and ML. Um, but yeah, I think we are getting there.

Taylor Houck: Yeah. And I'll tell you, this is something that I'm working on every single day, and that scoping is such an important part because you can't just roll out a blanket policy across the entire estate. Each application, each workload needs to be considered with its own unique considerations, for lack of better word,

Chitra Jadhav: Yep.

Taylor Houck: I mean, there are certain cases where I know at NBC Universal, if you've listened to the podcast, you've probably heard me talk about this before, but we ran a lot of SAP workloads.

Chitra Jadhav: Mm-hmm.

Taylor Houck: when you're running SAP on AWS, there's only certain instance types that are supported, and there's a requirement for how many [00:32:00] IOPS are needed on each of your storage volumes, and you must be compliant in order to get support from the vendor.

So the trusted advisor, the, let's say, recommendations out of traditional third-party FinOps tools did not consider those aspects, right? So then they were just inherently invalid. Now, that's when you need to roll out, you know, a, a, an aware tool or an aware-- I mean, in my case at the time, it was me personally doing the, the work, right?

Of considering these aspects and rolling it in. But, but you did touch on an important topic, Chitra, and we're coming to the end of the, the time that we have together today, but I need to hear your thoughts on AI because this is something that is taking our industry by storm and there's really two ways in which it's impacting FinOps.

One is pretty obvious, right? And it's that AI is a new category of cloud spend,

Chitra Jadhav: Yeah.

Taylor Houck: it is [00:33:00] one that is growing rapidly, right? Especially in this moment that we're in right now in the first half of twenty-twenty six.

Absolutely. Spending more on AI than they were planning to in the first part of twenty-twenty six.

And then the second aspect kind of plays into that because you can also use AI in doing your job better, right? We can be the ones consuming the tokens

Chitra Jadhav: Yep.

Taylor Houck: more effective. How are you seeing AI play out and what is your, your kinda take on where things are going?

Chitra Jadhav: so when AI started becoming a common word most recently, uh, there was a lot of panic as to what would happen to FinOps. But to

It's gonna be fun to

foundation of your regular FinOps, there's no way that you can succeed in a AI setup because without a proper governance, your CPU [00:34:00] to GPU transformation is going to be You, you need to know how you establish your cost estimations for the training models or for retraining them. Then there's gonna be a different template for your experimentation that requires so many times, uh, you know, repeating those costs, um, and maybe there'll just be wastage out there. And then the cost consumed for your AI inference finally.

So there are additional metrics that you'll start... you'll have to start tracking. The time to live becomes key metric, uh, because you don't want to be, um, turning something on and, uh, having a dev runbook running for a long, long time when you know that you're going to into prod after maybe six months down or four months down.

Now, um, so it's, it's a different ballgame, yes, but without having your basics sorted, you're going to fall flat because, um, that's the... You know, it's like having a version two of something when you've not even figured out a version one, right? So [00:35:00] that's the base foundation. That is what I would say. And yeah, you-- without a modern technology stack or without having a AI-compatible stack, there's no way that you, you can expect to smoothly fit in or keep pace with the rate at which the AI progression is happening.

So you have to be prepared for that.

Taylor Houck: see how it all plays out, isn't it? Um, awesome. Well, Chitra, this has been an incredible conversation. Before I let you go, I do want to, uh, to, to touch on a, a few more topics kind of outside the area of, of, uh, cloud cost management and optimization. Um, and one in particular is kind of the, the antithesis of, of AI.

I understand that you have been a, a mentor in your, your time and, uh, have done a number of different volunteering activities. Can you tell us a little bit about the, uh, the things that you do outside of work?

Chitra Jadhav: So outside of work, uh, I like, [00:36:00] um, networking here in Toronto. Like, uh, when I was in many of my organization, I always made sure that I participate in other events from a s-- you know, like, um, a charity point of view or raising, uh, awareness and social, uh, impact, um, clubs and associations. So that has always been, uh, a topic of interest.

I've also been part of Toastmasters club. I, I like public speaking. I like speaking in general.

Taylor Houck: Well, you're great at it, as evidenced by, uh, today's conversation.

Chitra Jadhav: Thank you so much. Um, yeah, so this is what I usually do, and, uh, I also plan to do much more in this space, um, now that, uh, communications have, uh, become so sorted because of YouTube. Uh, why not, right? Like, having more, uh, on-screen presence, that's where I would like to go.

Taylor Houck: Amazing. Well, I have really enjoyed today's conversation and, uh, thank you so much for coming on the show.

Chitra Jadhav: Thank you so much for having me, Taylor.

Taylor Houck: And thank you to our audience. If you got [00:37:00] something out of today's conversation, which I'm sure you have, please share this episode with someone who needs to hear it. This has been another amazing episode of FinOps in Action, and we'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, empowering teams to optimize cloud costs with deep detection remediation tools that actually drive action.