Episode 7: Interview with John Kodumal, CTO and Co-Founder of LaunchDarkly

August 17, 2017


Pramod HS

John Kodumal is CTO and co-founder of LaunchDarkly, a continuous delivery platform. John was a development manager at Atlassian, where he led engineering for the Atlassian Marketplace. Prior to that, he was an architect and advanced technology researcher at Coverity, where he worked on static and dynamic analysis algorithms. He has a Ph.D. from UC Berkeley in programming languages and type systems, and a BS from Harvey Mudd College. All in, he has over 15 years experience building tools for developers. He climbs rocks, ice, small boulders, and the occasional building, and enjoys juggling startup and family life.


Hello Everyone welcome to MappingTheJourney. I am Pramod. Before I start this show I want to share the theme of this episode. How can a simple idea turn into a successful product? Hear about the journey of a researcher, engineer turning to be an entrepreneur. You will also know about the experience of starting a startup in Bay area

Pramod: Welcome to the show

John: Thanks for having me

Pramod: John give us an introduction about yourself.

John: My name is John, I’m a CTO of LaunchDarkly. I have an interesting and long career, kind of in the tech space. I did an undergrad degree in computer science, went to Grad School did a Ph.D. in computer science and then have just really got attracted to developer tools and developer productivity. So I have worked at companies that were building things to help developers to be more productive at their jobs. I worked at a company called Coverity, which is a really interesting kind of like the technical product that detected bugs automatically. Really cool stuff about this and it worked off a lot of my research that I did in grad schools so it was really cool to productize some of the stuff that I do my grad work and I kind of shifted gears a little bit and I worked for Atlassian. I was at Atlassian for a little over five years and I think what I learned at Atlassian was a bit better how to manage people and how to get more involved more holistically instead of just being the individual contributor or an architect like I was at Coverity. Left Atlassian about three years ago and started LaunchDarkly.

Pramod: What kind of problems you enjoy solving the most?

John: Yeah like I touched on, the thing I’m always attracted to the problems developers have. I think one of my many strength and weaknesses at the same time as I can really only understand the problem at its heart if I have the first-hand experience with it, I can’t kind of look at a problem outside of my domain or expertise and really get a good insight into how to tackle that problem.

So with developer tools, I love programming, I love programming since I was a kid and so I look at problems I face as a developer when I write terrible code, you know I wish I had better tools. I could automate the detection of those problems and that’s why I went to Coverity. Now you know I’m looking at continuous delivery and need to deliver better software faster and LaunchDarkly is attempting to address some of those problems and we use LaunchDarkly ourselves we dog food quite extensively and I gotta say that for me it is an indispensable product and I’m super proud to have worked on into building it now.

Pramod: Explain LaunchDarkly to us?

John: LaunchDarkly is in simple terms, it’s a platform for feature flag management. So if you are not familiar with feature flagging, it’s a technique used in some of the best engineering teams like Netflix, Facebook etc they all use feature flag in their development practices. And the idea is to essentially put a conditional guard around new code that you introduce and have an external system like LaunchDarkly control who gets to see that flag or who get to see that new code path.

And the power of LaunchDarkly is that you can make that change instantaneously, in other words, you can decide to roll out a new code path or turn it off or kill switch it instantaneously without having to redeploy your software. So from DevOps perspective, it makes it incredibly easy to roll back problematic changes that have been pushed up to production without going through a cycle of rebuilding and redeploying software. And from that basic use case like my DevOps ops team gets to control change and kill switch bad change and ton of additional use cases things like AB testing control rollouts to user tests things like that.

Pramod: Feature Flag is a relatively simple idea. How did you convince your self that this idea can be made into a product?

John: Yeah I think as an entrepreneur that’s one of the first challenges you have is when you ideate, you look at every idea that comes along and your worst critic. You look at your idea and you think this is incredibly simple anybody could build this or you start googling around looking for other alternatives and you convince yourself that the market is saturated and there is no space for you out there and I went through a bunch of that with my co-founder Edith thinking through the different things that we might work on.

I think the biggest thing is, any problem when you think about it at scale can be an amazing set of challenges around it. As we dug into this feature flag concept we started recognizing what an incredibly rich area there was for the full-featured product. Like when we look at companies that are using home grown solutions. It turns out adoption isn’t all that great, it’s not what it could be and the problem is that there’s so much friction introduced by the process of using feature flags and managing them. And if you could take that away and make feature flag a much more pervasive part of the development practices and style then it just makes the dev organization that much faster and much more efficient.

Pramod: You always wanted to be an Entrepreneur or LaunchDarkly just happened?

John: It was a long process. My co-founder Edith, we had known each other since college. So we have known each other for quite a while, we never worked directly together but we were good friends and we had this basic trust in each other. I think around 2008 or 2009 we’re having a dinner party and we kind of did like a pinky swear that we were going to start a company together one day and it was a matter of timing. And the right timing came around 2014 for us where both of us were looking for changes in our career and what we were going to do next and it was the right time for us

Pramod: So you have the idea at hand, How did this idea turn into a product? Tell us the journey

John: Yeah there is a lot of aspects to that. I think the first one was, me as the technical lead I started building something just to see where it took me and we came up with a prototype that did something interesting within about a week. We started showing it to people and that’s one of the times where Edith’s kind of expertise and she’s an incredible networker she knows like everybody but was an opportunity to kinda show it around to dev teams in various spaces and see if anybody thought it was an interesting prototype and it turned out, it was to a lot of people and that combined with our understanding and we had seen this product being built before by internal tools teams and companies like Atlassian, Facebook. There’s a number of blog posts out there that kind of hinted the rich functionality with some of these really strong engineering teams have internally. We kind of felt like that was enough evidence that this might be something. So we really took off after that.

Pramod: Was there any push or any first customer that played a role?

John: It was tough, I mean the first customer is always the hardest and then getting the first customer to pay you anything is incredibly challenging and especially with what we were trying to do. We were running with like a very lean approach. So the first project we had I barely had anything, you could basically turn something on or off and I think I had like a slider you could do like a percentage roll out. But the basic architecture was there and we got a company to try us out and it was great to get feedback from someone using us in a production environment.

I remember the first day they turned us on just seeing some gauges light up and I think we’re doing about six requests per second. I thought it was insane because it was blowing my mind as everybody was using us at any kind of scale and now we are doing something like 50,000 requests per second. So things have grown tremendously and yeah that was how it started

Pramod: How easy or difficult was it find investors?

John: I think the experience differs pretty dramatically. if you have some track record history of being an entrepreneur and building something successful it’s too much different. For us even though we were kind of veterans so to speak in the industry we’ve been around for quite a while we were not like 23-year-old kids at school or something. it was still quite challenging, we weren’t really well known we didn’t have that track record of entrepreneurship and so a lot of it was I guess the term that we used back was team dream.

I was convincing people that Edith and I were the right team that our vision was good and we’re pretty good at convincing investors that the team was good, we have good pedigrees, I think we speak pretty eloquently about what we are trying to do.

The problem we had is that the vision wasn’t quite there yet, it took us quite a bit of iteration to really just like have the confidence to trust our own vision for what we wanted to be. We were getting a lot of feedback that was more like the AB testing side of things and when you start thinking about LaunchDarkly, I think people that are familiar with the competitive landscape immediately gravitate towards AB testing tools and things like optimize. And so articulating our story when some of our customers were saying, hey we really like you to just be optimizing for the backend which is not what we want to be. What we were working on was, we were focusing on a product was a completely different direction but we didn’t have the confidence in ourselves at the time to articulate that appropriately.

It took us some time to get a couple of customers and get the right feedback cycle where we found customers that understood our value proposition, maybe even better than we could articulate. It is something distinct from what those AB testing tools were doing. And once we were able to take their words and incorporate them into what we were thinking I really stick to our guns we put forth a much more coherent message and I think the investor’s kind of understood that and it became a little bit easier. I mean that’s kind of what the landscape looks like when you’re looking for a seed round you know pre-product, pre-revenue and kind of every stage along the way things change in the metrics the things that investors are looking for a change.

Pramod: Vision and goal are very important. And this evolved with your experience with the product.

John: Yeah absolutely I think if we didn’t have the experience base that we had if we didn’t have an understanding of how this could be used at scale, we wouldn’t have been able to articulate that story

Pramod: You have the idea and investors, bringing in right people is the next important thing. How did you manage to do it?

John: I think strong leaders, it’s a little less of a problem. if you treat people well throughout your career you end up being in a position where you know some of your network wants to follow you, they trust you and want to be a part of the things you’re doing next. That was what I really tapped into I think in my career I found a set of people that I knew I wanted to work with and you know they believed in me enough to come along and join me on this particular adventure.

For me, I think that that’s the path that I took but it can be extremely challenging outside of that, especially if you’re like pre-product trying to convince someone that what you are doing is interesting that they should care. For talented engineers as you are aware they have so many options these days. They could be working at any company making a really good salary and sacrificing some of that and accepting some of the uncertainty and stress of being in an early stage startup and frankly like small potatoes and not even having processing place is very hard.

I think that base of trust really helped with me because like our first engineer Patrick had a kid and for me I felt tremendous responsibility to not only make sure I was bringing him into a good situation but make sure that I was able to make sure that he could take care of his family and all that. So he trusted me, and I made sure that those important things were taken care of.

Pramod: Engineer vs CTO, did you have to make any fundamental changes in the way you think about the product?

John: Yeah, it was a very hard challenge for me, and I think I owe a lot of credit to my co-founder Edith. I think I did not really have a mature understanding how to take a product to market especially because of some of my histories like I was in companies that were just doing really really well Atlassian is an incredibly successful company, and so as an example I was part of the Atlassian marketplace team and we built the thing and people showed up and the company started doing transactions to the market place almost instantaneously. So I had this mistaken understanding that that’s how the world worked, you build a great product, you have a good engineering team, if it fits the need and people will show up, and it’s obviously very naive and that was beaten into me very quickly, and I think now I have a much more well-rounded picture of what it takes to take a product and put it out in front of people and really get people to care about what you’re doing when you’re in your early stages.

Pramod: If you have plans of starting your own startup, how important it is to understand the business aspects working at the large organization. Most engineers fail to do that. They will stick to their domain or engineering side of the product.

John: Yeah, I think absolutely. I think that and a lot of times engineers gravitate towards you know the things that they can impact most directly as an engineer and I think one of the challenges or one of the things that a good entrepreneur has to do is sort of divorce themselves from that and purposefully push themselves towards an aspect of the business that they didn’t really need to care about previously and I think absent that it’s very hard to see how you can succeed.

Pramod: Founded in 2014, are you happy with the growth?

John: Yeah, I’m astounded by the growth. I think it’s been a fantastic ride so far. What we had come to understand which we probably didn’t know about when we first started the company is that we are building a piece of mission critical infrastructure, so the trust and requirements the companies have when deciding to adopt our platform is that the bar is set pretty high.

We had this original idea that like feature flags is a service that we sell that for $29, and everybody will flock to you like the seltzer business that’s not the reality. The reality is that the problem that we were solving is much more difficult at scale and so the companies that want to buy us are much larger companies. The value that we provide those companies is much greater, and the trust that they need to have on our platform is pretty immense. So I’m happy that we are starting to see these large companies that are driving their businesses off of us. Actually just yesterday we had a quote from one of our customers said that LaunchDarkly is the one irreplaceable piece of their stack and this is a really large company. So that was tremendously flattering.

Pramod: Whats the scale at which business is running today?

John: It’s a pretty surprising scale. This is another thing that I hadn’t really considered is when a company really adopts us when they use us kind of pervasively throughout the development practices we become something that’s called all the time, it’s something you wouldn’t want to pay for us by call to us as it just wouldn’t make sense for us. We see actually like billions of calls for the request for future flags on a daily basis on hundreds of billions of events per month. We have hundreds of thousands of mobile devices connected to us at any given time and tens of thousands of servers connected to our platform at any given time.

Pramod: Just curious why mobile devices?

John: There’re customers that are deploying us within their mobile application so we can actually do feature flagging on mobile devices. So companies like GoPro if you are using GoPro mobile app, it is actually LaunchDarkly that controls some of the functionality you see in that mobile application. So many of their customers are connecting to us and through that.

Pramod: Whats next step for launch-darkly?

John: You know I think for us growing as a team and becoming a little bit more mature as an organization is kind of the next step. We have found our home here. We are based in Oakland now, and we moved from an incubator called Heavybit in the city. So we are kind of just stepping out on our own at this point and building our own culture and building a sustainable business. I know that’s super vague, but that is kind of what we are trying to work on next

Pramod: If you could time travel back to day one of your startup and have 15min with your former self to communicate any lessons you’ve acquired, what would you tell yourself?

John: I think one of the greatest lessons that I learned especially after we started seeing larger customers using us and depending on us is, I think starting out a company you’d think that the hardest challenges are technical and you are going to be working long hours just pushing out more and more products.

And I think a lot of the stress that I actually see is because you know I have such ownership over the product and the company that I can’t ever turn it off. So I’m always at some level thinking about it, and there’s no way to shut that switch off, and that can be incredibly challenging for your personal life. We are pretty good about work life balance here and my work hours are not insane. I spent a lot of time with my family, but there’s a part of my habit that is thinking about the challenges and problems that we’re facing on LaunchDarkly and so if I could time travel back I guess I would warn myself that this was going to be a thing and probably give myself some pointers on how to control that a little bit better and be more present with my family when I’m at home and more present at work when I’m at work.

Pramod: To what do you attribute your success?

John: Yeah, I’m trying to answer that in a very generic fashion. I think at some level probably the biggest factor is choosing the right people to work with, bringing the right engineers with the right attitude on board and having a great co-founder. I’ve seen so many startups fail because essentially it just turns into a bunch of teeth smashing where engineers don’t trust each other, the team doesn’t trust each other, or the co-founders don’t trust each other. We have had relatively few of those challenges this year especially like my relationship with my co-founder is really good. I mean like we yell at each other all the time but it’s sort of like healthy yelling where you know underneath it all you know that you have each other’s best interest at heart and the company’s best interest at heart and you’re both just trying to communicate how you think you need to get there or whatever. Arguments come up, but at the same time, you’re on the same page and the same team.

Pramod: What is the most unpopular opinion you have on entrepreneurship?

John: I think when you start a company you have this assumption that because you are the founder of the company you own it and you can take it in the direction that you want to take it and you are your own boss, and you’re in charge of everything. And the reality is that you have much less control over that than you might think. In several different ways, I think the most commonly you know mentioned piece of the data point on this is like okay when you get investors, and you have a board then you have a boss. You know your board is part of that equation as you have to take their input into account. But it goes much deeper than that when it comes to shaping your business external things that are outside of your control will take you in a direction that might not be what you anticipated, and I think it’s important to understand that and work with that.

As an example the market might be taking you down in a particular direction, might be taking your company down like an enterprise sales path. And you might not have envisioned that when you started your company and you really really wanted to build a SAAS business and focus on self-serve customers. If the market tells you one thing and it’s explicitly telling you that this other business model that you had in mind is not going to work, you may be the boss but you have to do that you have to take that other path because that’s what the market is telling and you have to build your organization structured in that way. I mean ultimately the company vision and what it becomes is still a product of you, but it’s a product review in addition to a bunch of other factors that aren’t necessarily directly under your control.

Pramod: Some of the initial challenges?

John: Yes, all the time, yeah it is that cliché of startup life being a roller coaster where you close a deal, and you think that everything is great and then you have an outage in something of value the trough again. What really gets you through that is just a belief in your in your vision and the help of your co-founder I guess. So having that strength just really helped to push through those hard times.

Pramod: Did you ever consider giving up or perhaps taking a job during this time?

John: No, I think we are in it for the long haul now.

Pramod: If you had one piece of advice to someone just starting out, what would it be?

John: As an entrepreneur, I think my first piece of advice is to get that first customer and get a first paying customer and do that as quickly as you can to validate whatever it is you’re doing. It’s probably not the most insightful thing to say, but you know going from 0 to 1 paying customer, I feel like is just as hard as going from 1 to 10 or maybe even 10 to 100. These massive inflection points indicate that you’re doing something right and until you have that first data point you don’t really know anything. So we pushed hard to get that initial validation “will anybody pay anything for this?” Once we had some of that validation, it gives us the confidence to kind of look for the next validation point and aims for that. So going from okay now we have somebody paying us on a monthly basis. I think our first customer was paying us nine dollars a month and it was more a courtesy nine dollars, but it was a piece of data you know. Somebody was willing to talk us into production and go through the hassle of writing us a check. We grew from that into you know a thousand dollar month deal, and we had other inflection points along the way that really got us into what we are doing.

Pramod: Thank you for your time. It’s wonderful talking to you John.

John: Yeah thank you very much.

Leave a Reply

Be the First to Comment!

Notify of