Platform as a Product - When Developers are your customers
Shuchi Mittal18:082023
Some of Fiserv's developers didn't know their platform team existed
Developers at Fiserv, it turns out, have been engineering platforms themselves without knowing an internal team already runs them. That surprise exposes the cost of project-by-project infrastructure support: bottlenecks, engineers stretched across 20 projects, burnout and a growing pile of one-offs. Shuchi Mittal, Global Head of Digital Transformation Platforms at Fiserv, traces how her organization set out to change that. What happens when platform engineers start treating developers as customers?
See the staged path from project mode to roadmaps, personas and self-service platform APIs
Shuchi Mittal, Global Head of Digital Transformation Platforms at Fiserv, explains how Fiserv's platform engineering organization moved from project-driven support to a product mindset that treats internal developers as customers. Drawing on Team Topologies, she walks through the stages of that shift: visible work tracks, product owners, roadmaps, feedback loops, user personas and an API-first, self-service platform.
“They were very surprised. So not only were they running into issues trying to use some of the platforms we have, but some development teams didn't even know we existed or we did some of the work for them.”
“So they're not seeing multiple platforms, they're seeing one platform.”
What you will learn
Map your organization to Team Topologies stream-aligned and platform teams to cut developers' cognitive load.
Spot the signs that project-based platform support has become a bottleneck: overloaded engineers, context switching and solutions trapped inside single projects.
Plan a staged move to a product model: a single visible work track, product owners, an idea portal, published roadmaps and quarterly planning.
Build feedback loops with developer customers through office hours, newsletters and a developer community.
Design self-service by defining user personas, mapping journeys across platforms and taking an API-first approach.
Key takeaways from the speaker
When Platform Engineering teams take ownership of specific parts of their platforms and create internally visible brand identities, they can articulate the benefits of their work to their customers - App Developers.
Explicitly define Product management responsibilities that include continuous improvement activities, such as road mapping, usage metric data-gathering, and quality monitoring. Focus on platform self-service to remove friction.
Details
Level
Intermediate
Format
Talk
Technical depth
Introductory technical
For
People manager, Senior individual contributor, Executive
Length
18:08
Published
June 2023
Questions this talk answers
Why does project-based platform engineering stop scaling?
No platform team has enough people for every project, so it becomes a bottleneck, and solutions built for one project stay invisible to the rest of the enterprise. Engineers spread across as many as 20 projects end up context switching and burning out.
How does Team Topologies help platform teams?
It separates stream-aligned teams that drive business value from platform teams that handle lower-order concerns such as VMs and monitoring, reducing cognitive load. The interaction model moves from facilitation towards offering capabilities as a service.
What practical steps moved Fiserv from project mode to product mode?
Teams first built a single work track to make all work visible, then recognized the need for product owners, collected requests in an idea portal, published roadmaps and started quarterly planning. They also tracked capacity for feature work versus firefighting and sized features in story points.
How can platform teams get better feedback from developers?
Fiserv started office hours, newsletters about what platform engineering was working on and its impact, and evangelism through a developer community of more than 2,000 people.
What comes after building individual platform products?
Recognizing that developers see one platform, Fiserv is defining user personas, mapping user journeys and pain points across platforms, and moving to an API-first, touchless model where developers wire platform APIs from a portal into their pipelines.
About this talk
“Infrastructure teams slow us down”, “It takes too many tickets to get my infrastructure”, “I have no idea what platform I can use or when” - Do these sound familiar? Your shared services organization is busy just trying to keep the lights on while the product development teams find it difficult to use those services and look for workarounds. This escalates into lots of one-offs over time and a mountain of tech debt.
Fiserv Platform Engineering organization adopted a product mindset for delivering platform capabilities a few years ago, which has helped us align better with the needs of our product development teams and accelerate the value flow through the enterprise. This involved aligning teams to capabilities, embracing agile development, user research and feature analysis, creating self-service pathways, defining and measuring KPIs, setting up feedback loops and so on. Terms like "customer", "value" and "friction" have become part of our vernacular. The tide is beginning to turn to healthy collaboration and dialogue between developers and platform engineering teams, and overall better outcomes for the enterprise.
Developers didn't know the platform team existed (73 s)
Why project mode stops scaling (87 s)
Developers see one platform, not many (69 s)
This is Shuchi. Hi, this is Shuchi Mittal at Fiserv. Welcome to the session and about how we and Fiserv have moved from a project-driven platform engineering approach to a product engineering approach. Focusing more on our customers, which are developers, internal developers. So we're going to start this off with a little poll. And I might not be able to see all the hands, but I would like to know how many, and we could just do this as an exercise. Raise your hand if you're a developer.
Focusing more in your organization and you rely on tools, platforms maintained by another team within your company to develop, deploy, run, monitor your applications. All right. I'm not sure I can see all the hands being raised, but I'm assuming there are quite a few here. So if that's the case, if you are one of those developers, let me ask you if this looks familiar, if this sounds familiar to you.
And one of the one that is most, when I was most surprised by or found the most interesting is feedback from developers and our organization. They realized there is a team that actually does manage one of the platforms that they have been trying to go engineer themselves. They were very surprised. So not only were they running into issues trying to use some of the platforms we have, but some development teams didn't even know we existed or we did some of the work for them.
And that was an aha moment for not only the development teams, but folks within platform engineering. So I don't know how many of you are familiar with team topologies. But one of the this has been a framework we we've been studying for quite a while and have kind of been trying to understand map out how this applies to us.
So the the the few constructs I'll just go over quickly because they're important to understand how where we are and how we've shaped our journey is the concept of these yellow lines, which are teams that are generating. Basically, basically your development, basically your development team, product development teams are called steam stream aligned teams, they're generating value value, value stream goes back to that, and they are.
They are the ones driving business value outcomes. And there are teams that are the the blue team here, which is actually a platform team trying to to do things so that these to add value to these stream aligned teams so that those teams can move fast and take care of higher order problems than the lower order problems.
Like, hey, I want to create a VM for my application, I want to figure out how to monitor my database. Those kinds of things can be handled up down here, then up into these individual teams, because there's the and the the theory is all about reducing cognitive load. And then up into this, you know, and I want to make sure that we're going to make sure we're going to make sure you're going to make sure we're going to work on those stream aligned teams that have to not only address end user problems or market changes, but now also have to figure out how to do things at a lower level in in the infrastructure.
And so that's kind of the concept around platform teams and stream aligned. And then there are other teams that, you know, complex subsystem teams and enablement teams that I don't get into. We don't have time for that. But the interaction models. What do you offer as a service? Where do you spend time actually sitting down and facilitating some of the work that these stream aligned teams are doing?
And so the the the the collaboration models that we have focused on are essentially how much starting with more facilitation moving towards as a service. So this is where the shift from project to product starts to happen. So kind of starting, you know, if you Google platform as a product, there are lots of commentary that comes up.
And I think this this is the one that resonates the most with me and my teams in platforms with a product mindset focus on higher order. We're solving those common denominator issues so that the development teams can focus internal development teams can focus on higher order problems to solve, because that's that's where we get revenue. That's that's what makes the business money.
So we need to focus on let those teams focus on those concerns that then those lower order concerns that I talked about earlier. So if I again, this is another representation of what we just talked about in team topologies focus on is this is where the FISA platform team sits. We have BU app dev teams, but we also have SRE teams that are focused on keeping the lights on, making sure those platforms are being set up in production for high availability, reliability from an application side.
Not from the platform side, not from the platform side, not from the platform side, we're deploying those applications to be highly available, highly reliable and managing the day to day operations around that. But they are also relying on the platform team to take those underpinnings and make them highly available, highly reliable and secure. And so taking that next to what what is the platform engineering team and at this point, you're asking what is what platforms are you talking about?
Well, we, we in platform engineering focus on making sure that the most used technologies are made available to our developers in a cost effective, robust, incredibly secure. And flexible and flexible manner. And those platforms could be very, I will just talk about a few. We have a cloud infrastructure team. We have our on prem private cloud infrastructure team.
We have container platform teams, API management teams, event streaming platform team. And so we have many of these shared platforms that many development teams use that we see value in managing the iterate and managing and iterating on continuously. And bringing that product mindset. So I've been talking about this product mindset. Right. And what is what is it that I'm actually getting to is the thought about focusing on developer experience.
So when we initially talked about that at Pfizer, it was, it was a lot of words. What does it actually mean? What does it mean for developers? What does it mean for the engineering teams themselves? Do I really need to change what I do from a day to day basis? And that's that caused the process to step back and say, well, where are we?
And how do we move into that model? And where we were at that point, this is about three to four years back. We were in a project mode. A development team would come to, you know, infrastructure teams and say, I am building this capability. I'm building this product. I need to expose APIs. I need infrastructure. I need VMS for my, on my data, the data centers, or I need to use this cloud capability or public cloud provider.
And we would have people engage with that business unit team on a, on a project level. So project would get spun up project manager assigned, uh, people from the infrastructure teams start to work with the development teams on all of those aspects that they are looking to for, for us to help with. The problem is at some point it's not scalable.
We don't have no, no, no platform team had enough resources to give to all projects. So we end up becoming a bottleneck, but also we don't see the, the effect that the solutions being effective because they are scoped at a project level. So if you created a new way to maybe an automation for, uh, using a cloud service, but that may be scope to that project may not be visible across the enterprise.
And, and, and I say that with a bit of, uh, you know, great assault here because there, there is still a platform engineering team and play, and they are sharing information across. And there were lots of efforts to bring people across those platform engineering teams to, to start talking about what they were working on. But it was always, Hey, we did this and we were thinking of doing this, but not concrete enough in terms of here's the timeline.
Here's what the feature set will be. And here are the teams that are depending on it, uh, from a business, uh, point of view or internal platform teams that were dependent on other platform teams. So it was very much, uh, uh, just a normal organic information sharing. Um, very quickly realized that since we are having a bit of a problem with one platform engineering, uh, person or engineer looking at 20 projects, uh, that they are participating in and really multitasking or context switching.
So often, but also just burnout, right. So kind of went to the next stage, which kind of makes sense from a change perspective to be smaller step is let's, let's build a single work track. Let's at least make all the work visible and start reviewing that as a team, as a, as project product, uh, a platform team rather than a project team.
And this, this, this is where this took us a while, uh, but three to six months, uh, depending on which platform engineering team we were talking about, but it also helped kind of elevate some of the roles we were seeing. Oh, somebody needed to talk about what we should be. What are the, what are the capabilities we're actually building across all these projects?
And do we need, we need a product owner, product manager. Um, and so we started having those conversations and that's where we, we actually recognize we needed a product owner. We needed somebody to, uh, uh, front end the conversation of what we should be building. And that got us to the next stage where we started publishing roadmaps and actually started doing quarterly planning.
What are the requests that have come in for this platform? You know, let's talk about event streaming or, or cloud platform. What, what are the requests that we're getting and how do we, uh, take that prior, uh, not, not even just prioritize. Initially, let's collect them in an idea portal and then start defining that, you know, what, what is this feature?
Get them into a roadmap, which we think makes sense. What we can do based on what we know at this point and what are the ads and start communicating that roadmap. And that, that actually started, uh, helping us with what our identity also, what are we doing? What is it that we creating quarter after quarter that lets us to the next thing?
Oh, we're creating this. What is the impact we are having? How are we doing with the, uh, with, on the operational metrics around, uh, running this platform? Um, you know, are there, are we spending too much time in, uh, troubleshooting and operational issues, or are we actually working on creating new features? Because at some point, if you're not spending too much time, you know, firefighting, you don't have that much capacity to actually work to feature work.
Sounds familiar to our developers, but it wasn't as evident to all of the, to all of us at this point. And so we kind of shifted towards understanding what our capacity was to work, do feature work, what our sizing, what is, how much is this feature in story points? And, and so that kind of led to a better, uh, viewpoint on prioritization.
Also, uh, we realized that too, for us to, uh, have a better, have better feedback from our, uh, consumers or customers, which are developers, we needed to get out and talk to them more. So that resulted in things like office hours or platform engineering, a lot of newsletters started publishing to say, what were we working on? What, and what impact we were having?
And as well as just evangelism through any community, we created a develop, we have a developer community, which is 2000 plus people strong that and growing every day to talk about what we're building on. So here's the next phase we are stepping into now. We're realizing we have all these platforms, but the developer sees them as one black box or not black box, but one thing.
And they want to be able to use that platform to build their application. So they're not seeing multiple platforms, they're seeing one platform. And that helped kind of starting to focus us on what is the user journey across these different, if I'm a developer trying to migrate my application into public cloud in a public cloud provider, what are the things that I may have to do?
And that touches many platforms today within platform engineering. So we've done the work on defining user personas, which is more than developers operations, as well as it's not the same. All developers are not the same. There's some that are working on public cloud migration. There are some that are working on just deploying and on-prem applications, mainframe applications.
So they're the different problem sets they have. So we've kind of defined the ones that we think are the more, are the real customers of our platform. We are mapping the user journeys and identifying pain points, and then starting to think about how do we take an API first approach to our platform? How do we make this touchless?
So you, you as a developer don't have to talk to so many people. You are able to go to maybe a portal, look at all the platform APIs, wire them into your pipelines and your, or in your applications and start using them. And we gather more, more metrics on usage and effectiveness from there on. So this is the, this is the stage we are at right now.
And I know I'm almost at the end of my time at this. So this, this has been really exciting for our, our teams. That's, it's a lot of change for a lot of teams, which are focused on basically making sure the platform was engineered in a way that was secure and operational. And we manage the operational metrics around it, but how this is our shift to managing the developer experience.
And basically that's the main thing. How we are shifting from engineering focus to developer experience focus. It through, through this entire journey. I know there'll be a few questions and I'm, I'm looking forward to hearing about those. Coordinators, I don't know if you switch to questions now or later. And I know there'll be a few questions. I know there'll be a few questions.
Transcribed automatically; names checked against the speaker record. Select a time to jump there.
Article based on this talk
What changes when platform teams treat developers as customers
Shuchi Mittal explains how Fiserv's platform engineering moved from project tickets to a product mindset built around developer experience.
In short
Platform engineering scales when it stops working project by project and starts running its platforms as products for internal developers. Shuchi Mittal, Global Head of Digital Transformation Platforms at Fiserv, describes how her organization made that shift in stages: visible work, product owners, roadmaps, feedback loops and, now, persona-driven, API-first self-service. The underlying change is cultural, from an engineering focus to a developer experience focus.
Key takeaways
When platform engineering teams take ownership of specific parts of their platforms and create internally visible brand identities, they can articulate the benefits of their work to their customers: app developers.
Explicitly define product management responsibilities that include continuous improvement activities, such as roadmapping, usage metric data-gathering and quality monitoring. Focus on platform self-service to remove friction.
The moment developers discovered the platform team
Shuchi opens by asking developers in the audience whether they rely on tools and platforms maintained by another team to develop, deploy, run and monitor their applications. She then shares the piece of developer feedback that surprised her most: some teams had been trying to engineer platforms themselves, unaware that an internal team already managed them.
"They were very surprised. So not only were they running into issues trying to use some of the platforms we have, but some development teams didn't even know we existed or we did some of the work for them." (1:44)
She calls this an aha moment, not only for the development teams but for platform engineering itself. It set the stage for asking what a product mindset would really mean.
Borrowing a lens from Team Topologies
To frame the shift, Shuchi draws on Team Topologies, a framework her teams had studied for some time. Product development teams are stream-aligned teams that generate value. "They are the ones driving business value outcomes." (3:27) Platform teams exist to take lower-order problems, such as creating a VM or working out how to monitor a database, off those teams' plates so they can move fast. The theory, she explains, is all about reducing cognitive load.
The interaction model matters too. Fiserv's collaboration has moved from more facilitation towards offering capabilities as a service, and that is where the move from project to product begins. Platform engineering at Fiserv covers widely used technologies, including public and on-premises private cloud infrastructure, container platforms, API management and event streaming, which it aims to provide in a cost-effective, robust, secure and flexible way.
Why project mode stopped scaling
About three to four years earlier, Shuchi recalls, platform work ran in project mode. A development team would ask infrastructure teams for APIs, VMs or cloud capabilities, a project would be spun up, a project manager assigned and platform engineers pulled in.
"The problem is at some point it's not scalable." (9:28)
No platform team had enough people for every project, so platform engineering became a bottleneck. Solutions such as a new automation for a cloud service stayed scoped to one project and invisible to the rest of the enterprise. Information sharing happened, but organically, without concrete timelines, feature sets or a clear view of dependent teams. One engineer might be involved in 20 projects, which meant constant context switching and burnout.
From a single work track to roadmaps
The first, smaller step was to build a single work track per platform and review all the work as a platform team rather than a project team. That took three to six months depending on the team, and it surfaced a missing role: someone to lead the conversation about what to build. Teams recognized they needed product owners.
"And that got us to the next stage where we started publishing roadmaps and actually started doing quarterly planning." (12:12)
Requests were collected in an idea portal, defined as features and placed on a roadmap that was shared openly. This gave teams an identity and made them ask harder questions about impact and operational metrics, including how much time went to firefighting versus feature work. Understanding capacity and sizing features in story points led to better prioritization.
To hear from their customers, platform teams started office hours and newsletters about what they were working on and the impact they were having, and evangelized through a developer community that Shuchi says is more than 2,000 people strong.
One platform, seen through the developer's eyes
The current phase starts from a simple insight about how developers experience all of this. "So they're not seeing multiple platforms, they're seeing one platform." (15:11)
A developer migrating an application to a public cloud provider touches many platforms along the way. So Fiserv has defined user personas (recognizing that not all developers are the same, from cloud migration to on-premises and mainframe work), is mapping user journeys and pain points, and is pursuing an API-first approach. The goal is touchless self-service: a developer goes to a portal, finds the platform APIs and wires them into pipelines or applications without talking to many people, while the platform teams gather usage and effectiveness metrics.
Shuchi acknowledges that this is a lot of change for teams used to focusing on secure, operational engineering. Her summary of the whole journey: "How we are shifting from engineering focus to developer experience focus." (17:17)
Watch the talk
Platform as a Product - When Developers are your customers, Shuchi Mittal, 18 minutes: https://www.womentech.net/video/platform-product-when-developers-are-your-customers
About the speaker
Shuchi Mittal is Global Head of Digital Transformation Platforms, leading the vision, strategy and delivery of technology that powers Fiserv digital finance products through APIs, event-driven architecture, data cloud and cloud reference architectures. She has over 30 years of experience across product engineering, technology strategy, reliability engineering, technical solution consulting and process transformation, and is a member of the Google Apigee tech advisory council and a Fiserv Expert Engineer.
Questions this talk answers
Why does project-based platform engineering stop scaling?
How does Team Topologies help platform teams?
What practical steps moved Fiserv from project mode to product mode?
How can platform teams get better feedback from developers?
What comes after building individual platform products?
Drafted with AI from the talk transcript. Editor review pending.
Frameworks and ideas mentioned
Team Topologies
Platform as a product
API-first approach
Story points
Cite this talk
Shuchi Mittal. "Platform as a Product - When Developers are your customers." WomenTech Network, 2023. https://www.womentech.net/video/platform-product-when-developers-are-your-customers