I work on the platform engineering team at Intuit. This post is reposted, with light edits, from the Intuit engineering blog — co-written with my colleague Avni Sharma.
Hey, Let’s Talk About Kubernetes#
If you’re a software engineer with a few years under your belt, you know Kubernetes is the de facto king of container orchestration. It’s powerful, it’s scalable, and it’s the foundation of modern cloud-native infrastructure. But let’s be honest—it can also be a beast.
At Intuit, we operate at a massive scale. We’re talking about a platform that serves 100 million customers across products like TurboTax and QuickBooks, running on an infrastructure with over 2,500 production services and 300+ Kubernetes clusters. At that scale, the complexities of Kubernetes don’t just add up; they multiply.
This work builds directly on a vision we laid out back in 2022, when we first described our plan to unlock the next 6X in development velocity through application abstraction. What started as a roadmap is now a production reality — IKS AIR.
The Outer Loop Trap#
When we looked at our development lifecycle, we found that engineers were spending about 42% of their time on “inner loop” activities—the good stuff like coding, testing, and debugging. However, a whopping 30-40% of their time was consumed by the “outer loop”: deploying, operating, monitoring, and managing infrastructure. (This split is consistent with what we’ve seen at peer companies — see Bain & Company’s 2024 Technology Report.)
True to our philosophy of “falling in love with the problem, not the solution,” we started by listening to developers before designing anything.
That outer loop shows up as wrestling with dozens of YAML files just to get an app running and a learning curve that can take a developer 30 to 60 days to get comfortable with the infrastructure. It also shows up as risk: misconfigurations that lead to policy violations or production incidents. The numbers backed this up: one in five support requests we fielded were about Kubernetes basics, not application issues, and configuration complexity was behind thousands of policy violations across our fleet.
Put simply: when developers spend a third of their time on infrastructure instead of features, that’s not a training problem — it’s a platform problem.
We knew we had to flip the script. Our goal became clear: empower our developers to focus on writing code and shipping it fast, getting them back to the inner loop where they create the most value.
The Vision: An AI-Native Abstracted Platform#
IKS AIR — where IKS is Intuit Kubernetes Service, and AIR stands for AI Runtime — is an AI-powered, abstracted runtime that sits on top of it. The core idea is to translate simple application needs into complex platform means.
Instead of forcing every developer to become a Kubernetes guru, the platform provides an opinionated, paved road that handles the heavy lifting. It’s about creating an interface where a developer can state their application’s intent, and the platform takes care of the rest—from configuration and traffic management to autoscaling and operational safety.
The result? Each service that moves to IKS AIR saves an estimated 24.7 engineering days per year in ongoing outer-loop work — paying back the roughly four days it takes to migrate in just a few weeks, then compounding as a recurring efficiency gain for the life of the service.
In the next part of this series, we’ll dive into the first pillar of how we accomplished this: abstracted compute and AI-powered autoscaling. Stay tuned.
If you’re curious what else Intuit’s engineering org is building, check out the Intuit engineering blog or their software engineering careers page.
