Guest: Billy Thompson (LinkedIn)
Company: Akamai
Show Name: KubeStruck
Topic: Kubernetes

Platform engineering promises consistency and speed, but most teams underestimate how complex it is to build a production-ready internal developer platform (IDP). At KubeCon in Atlanta, Billy Thompson from Akamai put that complexity front and center through a live, high-pressure, 90-minute workshop challenge.

At KubeCon, Thompson, Senior Global DevOps & Platform Engineering, Office of the CTO at Akamai, helped run one of the most unique technical workshops of the event: a Capture-the-Flag style exercise where teams had just 90 minutes to build a production-ready IDP on Kubernetes. The goal wasn’t speed for the sake of entertainment. It was to reveal the reality behind the ongoing “build vs. buy” conversation in platform engineering.

Thompson has seen a consistent trend. Teams often believe they can build a fully featured IDP themselves because Kubernetes appears flexible, powerful, and extensible. But as he pointed out, the real work sits in the layers beneath, where decisions compound and complexity multiplies. The workshop was designed to surface that complexity in the most direct way possible.

Even experienced engineers were surprised by how long certain tasks actually take. Thompson highlighted network policies as a simple example: it’s not just about writing the configuration. Teams must decide what their default policies should be, which rules apply cluster-wide, where exceptions belong, how namespaces factor in, and how service mesh behavior fits around those decisions. “It’s a whole logical thought process,” he said, and in a real-world environment, that planning takes far more time than expected.

Throughout the workshop, participants teamed up—some with colleagues, some with total strangers. And while the event had a competitive element, it also created a collaborative environment that resembled the cloud native community itself. Engineers may work for competing companies, but they openly share ideas, patterns, and best practices across projects and conferences. Thompson noted that the workshop unintentionally showed both sides: competition within the challenge, and cooperation through shared learning.

He also saw a recurring pattern that aligns with what many platform engineering teams face: even advanced practitioners hit bottlenecks. Every table eventually raised its hand for hints from the LearnK8s facilitators. That was a positive sign in Thompson’s view. It meant the challenge design struck the right balance—technical enough to stretch everyone, but accessible enough to keep teams engaged and determined to push through.

But the workshop wasn’t only about exposing friction. Thompson wanted participants to walk away with a clearer path forward. He demonstrated how reusable “golden path” templates can prevent platform engineering initiatives from spiraling into endless technical debt. Instead of reinventing configurations for every new service, teams can build once, standardize, and iterate sensibly. In fact, he noted that many organizations could realistically operate with a single golden path template for their IDP. To reinforce this point, he closed the workshop by showcasing a fully open source, Apache 2–licensed CNCF project that embodies this template-driven approach.

When asked whether the workshop was meant to showcase Akamai’s approach or offer a learning experience, Thompson shared two perspectives. From the company’s standpoint, it was an opportunity to demonstrate a working, open source, production-ready IDP that teams can adopt or study. But personally, he wanted to highlight something deeper: the overlooked problem of complexity. He believes many platform engineering failures stem not from hype or misalignment, but from teams underestimating how complicated it is to build scalable, multi-team platforms that deliver a good developer experience.

He pointed to well-known industry statistics claiming that more than half of platform engineering initiatives struggle or fail. While some attribute this to architecture missteps or shifting definitions, Thompson believes the real issue is the lack of a grounded approach. Teams jump in without fully understanding what they’re up against. His workshop provided a condensed version of that reality—an accelerated environment where pressure exposes the decisions, trade-offs, and gaps that normally spread across months.

The strong response to the workshop has already shaped future plans. Thompson and the organizing team intend to bring the session back for the next KubeCon. They are also actively exploring virtual versions, larger formats, and even a global tour. He said early conversations have expanded from “let’s do it once” to “this should be an ongoing experience,” especially now that they understand how to improve the logistics and scale participation.

Looking ahead, Thompson is considering how future workshops might evolve. The 90-minute IDP challenge format works well, but the themes could expand to highlight new challenges emerging across cloud native teams. He sees opportunities to focus on governance, developer experience, sustainable platform operations, or other high-pressure problem domains. But at its core, he still wants the workshop to stay fun, fast, and grounded in real-world engineering.

As the interview closed, Thompson thanked the TFiR team for pushing him to think differently about the experience. He said the questions gave him new angles to evaluate and refine future workshops—part of the same iterative learning that platform engineering itself relies on.