Introduction

Han Yuan 0:00

Hey, welcome back to the show. Hitesh and I are stoked to present our second conversation with Bala Madhuri, someone I've had the privilege of knowing for over 17 years. Bala is a true pioneer in the data and AI space. He's been at eBay, JPMorgan Chase, and now Calendly. Today, he's serving as VP of Data Analytics and AI at Calendly. And broadly, he's building his career pushing boundaries in how companies leverage data and AI. I'm especially glad that he's making calendar management less terrible. Today's conversation is a scaling journey. Early eBay operated in what we consider the dark ages of computing infrastructure. Unlike today's world, where cloud resources are available to you with just a few clicks. And in most cases, even that is automated. eBay had to build and manage their own massive data centers. We're talking physical machines, real estate that they actually own. Bare metal approach meant that providing capacity was an incredibly delicate balancing act. You needed enough machines to handle traffic spikes during peak shopping seasons, but you couldn't afford to have too many expensive servers sitting around idle the rest of the year. The other thing we should talk about is that if anything blew up, people like Bala, and Bala in particular, was often on the hook. So how do you transition your career and background from a leader of foundation teams towards business teams. Paul's gonna tell us. He's gonna tell us how he transitioned from leading some of the most mission critical systems and teams at eBay including the database teams and the site reliability engineering teams, to spearheading the foundation for what would become eBay's modern AI infrastructure. We'll be diving into this journey, leading the AI teams during this critical period, exploring challenges, breakthroughs, lessons. So without further delay, let's do this.

Hitesh Chudasama 2:11

We're excited to

Bala's Background

Hitesh Chudasama 2:12

kick off this episode with Bala Madhuri, an engineering executive who has been part of the technology world for the past 20 years, working across companies like eBay, JPMorgan Chase, LTK, and now at Calendly. Han and I had the opportunity to work with Bala at eBay and see the awesome work that he had done. Bala is passionate about not just technology, but building high-performing teams and delivering creative technology solutions to drive value for customers and companies. Bala, welcome to the show.

Bala Meduri 2:44

Thank you, Han and Hitesh. Excited to be here. It's going to be fun.

Hitesh Chudasama 2:48

Bala, before we jump into your projects, can you give our listeners a brief overview of your career journey? You've held leadership roles at eBay, JPMorgan Chase, and more recently at Calendly. What led you to the tech and data arena?

Bala Meduri 3:03

After I came to the United States, I'm going 20 years back, I did my master's in engineering and data captured my imagination. I started my career in the data world. The more I worked in data, the deeper I got. It's been a fascinating journey because every end user experience, at the end of the day, there is data behind it. I worked in a lot of storage systems and transactional data platforms. When I I realized how it was impacting some of the core business and how it's adding business value. I felt like this is the place that I'm going to continue my career. I spent most of my career at eBay and I started off from a hands-on engineer in data to grew into the team lead and then to the managers and one built on top of each other. One of the things I realized was not to be afraid to jump into different things. Although it's pretty scary, but it's also important to understand like, hey, if I were to take a different role within the data or expanded data role, how would that play out, right? Always helped me to jump into a place where I feel very uncomfortable, but then you learn like there is no tomorrow and it always has helped me to continue in that journey. And so one led to the other, I expanded my roles and I jumped into different kinds of roles. And if you roll forward to now.

Han Yuan 4:36

Hey Bala, if I could interrupt real quick. How did you make the decision to go from an IC to a manager?

Transitioning into Management

Bala Meduri 4:44

That's a great question. It's one of the most, probably the toughest break, right, to get into that. And so here is how it transformed. There was a time where while I was a hands-on engineer, there used to be a lot of questions from our customers. Hey, how do you do this? Like, hey, we have a problem on this. So I became somehow a de facto person to respond to a lot of those customer requests. In fact, I felt compelled, like, hey, when somebody asking our team the questions. I could not not answer that. And I almost always was the first person to respond to the customers, right? And so what happened was, and I did that for a few months. And when a role... Our opportunity came along it became very natural who is the person to take that role because it just like how I was like Responding to the customers or was working with the team. How do you how do you how do you solve those requests that we were getting right? So that's what led me when there was a team lead and manager position. It was almost unspoken Who would be the person to do that? Maybe one lesson there is I would say the behaviors should come before for opportunities. So it's not that when there is an opportunity that, oh, okay, well, why are you the right person? You already demonstrated your behavior. So when the opportunity comes in, it's not even up for question. So you position yourself to get into that role. So that has really helped me. So that's how my transition from individual contributor to manager, it was scary, but it also felt a little bit natural because I was doing some more team lead, managing like stakeholders and things like that.

Han Yuan 6:34

Were you worried that you would lose some of your technical chops? And I only say that because I know you're badass. I've seen you. You've had many, many people. You were still on the keyboard. And so I know your answer, but I'm curious, when you made that decision, were you worried that somehow you're going to lose your technical stuff?

Bala Meduri 6:51

Yeah, I mean, initially, when I was transitioning into that role, it's not like I'm doing 100% people management. When I first transitioned to a line manager, so I was always a coach, but also pretty hands-on contributing. So So that it's almost like I'm part of the team, right? Contributing to the individual, the hands-on engineering pieces. And I never felt that it was a role that's 100% people management and you're hands off the key, right? And that helped like for a couple of years until I got into senior manager. So I was never worried because I was playing both to some extent for some time, but then there comes a time where it feels very natural. Hey, your priorities, when you're leading, for example, 50 people organization, you cannot be like spending more time on the hands on the keyboard, but like you have a bigger priorities on how do you manage the team? How do you get the highest, biggest outcomes from on the team and so on. So it kind of feels natural as you grow, you taper off a little bit on the hands-on stuff and the other, the core leadership things takes over. Either to this extent, I still like code and more like at a personal level, I don't want to be contributing in the production code, but I still do, right? So it's just fun playing with the notebooks and right, so, and it's become super easy as well, right, so.

Hitesh Chudasama 8:17

Thanks, Bala, for sharing the journey from IC to leadership role. I know it's pretty tough at times, especially when you are in different kind of meetings, talking about it from a business landscape perspective and strategy, and then going and being hands-on as well. So that's a great transition. I know one of the other transitions you talked about was at eBay, where you were leading the machine learning and AI research group. Can

Leading the Machine Learning and AI Research Teams at eBay

Hitesh Chudasama 8:42

you walk us through that transformational phase and what was the big picture behind creating a unified machine learning platform?

Bala Meduri 8:48

Yeah, so this was... I believe in around 2018, I was actually leading the SRE team, the Site Reliability Engineering. Think of it like global operations, technical operations for eBay, right? So in other words, anything that happens, bad experience for the customer, the bug stops with me. So, or from a technical, hey, but why? No pressure. Right, right, no pressure. No pressure. Woken up at 2 a.m. and you got 30 seconds to get your act together and to understand what's going on. But so from the SRE, I transitioned into the database platform or the AI ML platform and also the data platforms. The goal was to empower eBay's internal developer community to seamlessly build, test, deploy, and retrain machine learning models, right? And why was that important to create a unified platform was that it's a tough journey, right? And if each developer group is kind of doing that entire life cycle, that is to build, test, deploy, retrain all those models, it's almost like everybody's reinventing the same thing over and over again. And by the way, it's not very easy, right? Like for example, managing your own GPU cluster, right? And it's not just the technical system side of it, but then you need to have an entire ecosystem, right? So like frameworks, right? And workspaces, right? And data aspects of how do you stitch the entire life cycle together. So it was happening like in silos and while we were limping along, it was not moving fast enough. And at the same time, at that time, the context of what was happening at eBay was there's quite a bit of innovation that was happening, like, hey, from a business context, we need to build a lot of these merchandising recommendations. personal buyer personalization, right? Or seller price guidance or language translation and shipping estimates. So there's all kinds of these business problems that we're trying to solve. At the same time, there's also this developer silos from how you build those models and deploy. So that is where the kind of story came together, the recognition that we need to build a unified ML platform, machine learning platform. This is also the time around when AWS was releasing SageMaker. Facebook was releasing FB Learner or Uber was releasing Michelangelo, right? So I think this is back in 2008. This is like seven or eight years back, right? But that was the problem of the day. And we recognized that. And then that was the beginning of like why the recognition that we need to build a central platform. And so that's how it started.

Han Yuan 11:50

This is really incredible. So what I'm hearing from you is that you were part of the movement where you were actually democratizing all of the machine learning workloads across the company.

Bala Meduri 12:01

Our strategy was always focused on the most urgent and high impact features, demonstrating a continuous progress on the platform and maintaining regular communication with the stakeholders. So I think prioritization, what has the biggest impact and key communication with the stakeholders was such an important thing. And we can talk a little bit about the mechanics of how we did that.

Hitesh Chudasama 12:25

I think that'd be great. One of the things, I mean, I was at eBay for 14 years and I know that at that time eBay liked to build a lot of things in-house. So in this particular case, when you went into this organization leading it and you were talking about all the other solutions that were out there, mainly asking in regards to buy versus build. Was that something that was considered at that point within this group?

Bala Meduri 12:48

Yeah, we looked, right? And I think even at that time, there was a lot of, even in the industry, a lot of open sourcing constructs are happening. We certainly looked at buy versus build. And one of the context for eBay is also that we have a lot of pretty strong technical teams whether it's infrastructure and the platform. And of course, eBay has been doing machine learning way before this platform. So I think one recognition is, hey, we can leverage our own internal engineering strength But importantly, how do we unify the forces so that we contribute to a platform based on the strengths that we have and what we've already been doing? And then the question was, okay, how do we come together in an extraordinary way so that we are all building one product or one platform rather than being in silos?

Hitesh Chudasama 13:41

How did you quantify your outcomes, especially in regards to time to market for the AI models? What was the way you were able to measure success?

Bala Meduri 13:49

When we build the unified platform, that development and deployment, we reduced the time to market from months to mere days. And that was one big quantifiable thing that we were able to demonstrate, right? And perhaps more importantly is that we enhanced that productivity, but also fostered a collaborative environment, right? Where data scientists and engineers could work together effectively driving AI capabilities to newer heights. And so every time

How To Measure Success

Bala Meduri 14:19

a developer contributes immediately can be leveraged by the rest of the community in eBay. Creating that collaborative environment was a big step forward to evolve the platform much faster. So ultimately, again, the way I think about the technology issue, it's an enabler to either drive business or to provide a compelling magical experience to the customer. One of the things that we did leveraging the ML platform was eBay is known for the discovery. So the discovery of the massive selection of items that you wouldn't be able to find anywhere else. So the discovery was a key thing for eBay. And in order to provide the inventory for the discovery, we need to make the seller experience simpler because we cannot have a friction in the selling flow because that's how you get the inventory into the system. And so what we did was complete reimagine that selling experience. A seller can now just take a picture from the seller's camera and we do the magic of matching with our catalogs and so on. And then within the next click, this is the item that I'm trying to sell based on the picture. The next click, the item is on the platform, immediately visible to hundreds of millions of customers. So really transform this seller experience and really And where we used to have important drop-offs in the seller experience, that solved that because we made super simple the experience. But also the inventory, the quality of inventory really improved as well because we're doing the matching from the catalog and so on. It's not like a free text, five-page form. It could result in inconsistent experience. So we made the customer experience simple. We solved the business problem. And that was all possible based on the platforms that we build. I also want to say there's extraordinary product partners that we work with and the excellent executive leadership support to make this happen. But it was such a fantastic journey. There's such a collaborative team ecosystem that kind of brought all this together. That seller experience was considered to this day pretty magical. Of course, it has been evolving and so on, but that's the example of where tech truly enables the customer experience, but also adds a business value.

Han Yuan 16:39

I suspect though that there's there's a piece of the story that's decidedly you. It's easy to connect the dots looking backward, but at the time when you took over the organization, what was the groundwork that you laid so that all of these pieces would eventually fall into place so that you could succeed? And having been in an executive role myself, as well as Hitesh, I think one of the things that any executive has to think through is from the moment that you start the job, there's a clock ticking, literally, where where you have to make impact. So there must have been all these things going on in your head. What was your personal roadmap of Bala managing his career so that you could exit this thing doing a good job?

Bala Meduri 17:23

Yeah, there are certain organizational framework that we need to lay out, right? So it's a big ambition, right? So one of the things that we build is what we call initiative core group, like AI initiative core group, the ICT, where we have all these stakeholders. like one person it's we need to mobilize the organization and of course there's a commitment both from executive leadership as well and we have to do this right because if you don't have this right framework it's going to fall apart pretty quickly i think operationally on a day-to-day basis does take like okay who should be in the stakeholders like how do we what's the cadence that you meet and and so on so all building those operating rhythms the communication and the stakeholders and prioritization the fellowship programs that's a lot of grind grinding right to kind of bring that together because if we don't have that uh pretty soon and And one of the things we also have was we continuously demonstrated how the platform is evolving. So, and it is important so that the stakeholders understand the progress that we're making and hey, this is the model that's deployed using the platform. And by the way, guess what? You contributed to the particular feature. So all these stakeholders should also have the skin in the game to make that platform successful, right? And I think for me, because it drives the pace of innovation, it drives the competitive edge that eBay has, right? To move faster, right? to have this platform and It's a lot of pressure, but it's also what happens is like once you start demonstrating continuous progress like week over week and every two weeks and demonstrating the value, I think it builds the confidence. So that and then and then it becomes almost like folks don't use anything outside of the platform and then the momentum builds and it's been it's been communication is the key. At the end of the day, what I realized is that tech is one part of it, but collaboration is the key. Empathy matters because folks do come in from various perspectives and maybe their own fears and ambitions. I think we have to recognize that and see where they can effectively contribute. So collaboration, empathy, communication, and also prioritizing on few things that are impactful, but also execute with a sense of urgency and with quality. So those are, I mean, some of the things that I know are important and I try every day to operationalize that and carry that spirit forward.

Han Yuan 20:09

One of the things that I heard from you that I thought was fascinating was that there's a key playbook that you did, which is you're showing the work as it's happening and ostensibly in a way that the management team can understand, like the higher-ups. So there's a certain language that you're using to communicate upwards, perhaps different than what you're using to communicate with the community. And then the other playbook that I heard from you that I thought was interesting, and I think this is encapsulated in your concept of empathy, is you were also sharing the credit. So you were showing the work, sharing the credit, and you were doing all of this at a regular cadence so that you were driving the predictability, showing the work, sharing the credit, rinse and repeat. And I think that's absolutely fascinating.

Hitesh Chudasama 20:55

So Bala, you know, wrapping up on this journey, as you like reflect back on, I know

Delivering Impact In a High Expectation Role Quickly

Hitesh Chudasama 21:00

it's been many years now, but as you reflect back on it, how did this reshape your view of leadership? And also, if you had to go ahead and redo this similar kind of effort again, what would you change? I

Bala Meduri 21:12

think one of the things I realized, again, I probably have mentioned this before, at the end of the day, technology is an enabler. And so what helped, and in

Lessons Learned

Bala Meduri 21:25

fact, what I won't change is that never lose the eye on the ball on the customer impact or the experiences to the customers and the business impact that you're making. So because at the end of the day, that's what like tech is enabling. I think I'm trying to see like what I would change is that certainly technical problems are not that complex and we can work through all that. And what I always found out was how you align with the broader organizational context. That's super important. And getting that alignment and the buy-in is very important. That's one. Like if you don't have, it's what I I felt in general, if there is no executive buy-in, like when I say executive, like CTO or like even CEO, a lot of this will fall apart. I mean, it's a tough journey when you're trying to accomplish, but there's also, for some reason, like there is no executive support. I think building that alignment is important from an executive framework. And then of course, how you build the alignment with the stakeholders with across different groups is also important. So once you establish those two, even at the engineering level, I think it's important to have worked with the engineers themselves, right? Why that's important and kind of bring that buy-in. And I think we need to establish some framework. What's the... in what's in it for them, right? Because if you don't establish that, ultimately the engineers are the who are contributing to it, right? So I think you have to work through different levels, right? And different layers of the organization to get that alignment and the buy-in. And so those are the ones. I think if I were to take back, I would still say how you prioritize on few impactful things and executing that with such a sense of urgency and the quality matters. So because, okay, you have this great vision but you're all over the place and not diluting yourself and not accomplishing things. I think that's super important. That's why I think align with the stakeholders. What are those three or four things that are important? And then execute them flawlessly. And what I've seen in general, different organizations react to it differently. The sense of urgency means different things in different contexts and different teams. So to me, that is very important. And I think a lot of your learnings would be more understanding and the human level, like how to build that alignment and at various levels and delivering the value is super important. Fantastic.

Han Yuan 23:54

You mentioned alignment a couple of times, Bala. And I was wondering if you could give us some tips because some of the characters you've talked about, like the CEOs and CTOs of eBay, they're pretty daunting figures. They're famous. They're in the news. Like I can imagine you're just like, you know, one out of a hundred thousand people. How do you build those relationships or not come off just obliquely sort of building the relationship. What are like two or three tips that you would give somebody who's like, okay, I get the fact that I need the CEO to buy into my stuff, but how do I bridge that gap?

Bala Meduri 24:29

Yeah, I think what executives would love to know is how that's going to contribute to the business value. And for example, we laid out some of the problems. So, hey, here is like what's happening, like some of the siloed work and how that's not getting stitched together and Why is that a problem? Why is that causing a slower pace of innovation and so on? And that is why maybe we're not moving fast enough in certain areas like building the models and deploying those models or bringing those customer experiences. So I think from an executive level, it's all about what is the core problem that you're trying to solve? What is this size of the problem that you're trying to solve? And reimagine if we were to get to the other side, like if you have solved this problem, Like, how would that picture look like? I think we have to be able to paint that picture in almost, I would say, strictly in a business context. So, and you can even add some dollars and cents as well. So here is how much we are spending and here is how much it's taking to build the models and then kind of map it to the new world. I think kind of build that in a language that it resonates with the ELT or executive leadership team. I think that's important. And most of the discussions there would be certainly in the business context. Like CEO cache, like, okay, well, what does that mean? So to our customers, what does it mean to the top line or bottom line? So you need to be able to paint that picture. And one of the things that we did, I vividly remember, is we're painting some of these possibilities of what the customer experiences that we bring in, if we do XYZ, like whether it's platform or certain core AI features, whether it's in the vision, language, and so on. So I think it's also important in my mind to see what the possibilities are with the technology, but paint that picture in a business context. So I think that that has been super helpful because when, if you start doing deep tech and you were depending on like your audience, you will suddenly lose your ELT. If you go deep into the technology, but not framing the conversation in what does it mean to the customers? What does it mean to the business? So that has been like, I think super helpful by the not just at eBay. I've learned it's the same exact playbook. JP Morgan as well, or LTK and so on. It's ultimately, what does it mean to the business and customers? And you have to ground yourself on that.

Hitesh Chudasama 26:54

Great. Thanks, Bala. These are valuable insights. Thanks for providing that. As we wrap up, we have our audience. Some of them are emerging technology leaders. Some of them are in mid-level management. And going through their different phases, what advice would you give them based on your experience? both on the personal as well as professional perspective. I know on the personal side, you do the GI meds. We'd love to see if you could go ahead and talk briefly about that. But what are some of the advices that you would give to our audience?

Bala Meduri 27:25

Yeah, yeah. So, yeah, if I start over on the GI meds, that would take the entire podcast by itself. But I think... I have my sources of energy and everybody has like limited energy. So one to understand what your sources of energy, but also manage it productively. So my source of energy, because you briefly mentioned GI meds. So it's like live every day with gratitude. And some of the practices that I have in my daily stuff is intermittent fasting, meditation, exercise, diet, sleep. That's what GI meds start with. And so those at least I like to believe I believe it contributes great sources of my energy, just daily good habits, if you will. And then, of course, manage those energy sources productively. That's one. I also think that we don't want to sit on our laurels. And of course, you did X, Y, Z. Awesome. But then you continue to learn, you continue to grind it out, and you never stop demonstrating. And I strongly believe that to this day. So, I mean, whatever you accomplish, that's awesome. But the tech landscape is changing so much. so much. So how do you keep up through continuous learning and continue to demonstrate that value? You never stop that. So that's another piece to it. And again, understand that technology is just an enabler. And your core goal is how you make your customers productive and delivering business results. And you should never lose the eye on the ball on that. And then I think I talked, one of the things is that have clear alignment across teams and focus on a few things by asking what matters the most in the changing situations. Again, delivering that with a sense of urgency and collaboration so that everybody has the skin in the game, everybody wins. I think looping a lot of the stakeholders at different levels is super important.

Hitesh Chudasama 29:22

Thanks, Bala.

Wrap Up

Hitesh Chudasama 29:23

I really appreciate your time and the insights that you provided to our audience as well. Han, is there anything you want to add?

Han Yuan 29:29

Dude, Bala, hanging out with you has been so awesome. Awesome. You've been one of my favorite co-workers over the years. Like Chin Inspiration, just the number of jobs that you did at eBay and hearing behind the music of how you went from the SRE team to really in a transformative position that I think, me personally, really laid the foundation for all that innovation that we're now seeing at eBay in 2025 is really, really neat. So I thank you so much. This has been so awesome. Yeah,

Bala Meduri 30:05

thank you so much, Han and Hitesh. And it's been a fun conversation. And yeah, so thank you. All

Hitesh Chudasama 30:12

right. Have a good one. See you.