How to Create an Engaging Mlops Platform Using Kubernetes, Open Source and Golden Paths

Even with an explosion in tooling, building a scalable and reliable machine learning platform with open-source technology can be quite a challenge. But beside this technical puzzle lies an even bigger challenge: how do you engage your end-users and make sure they actually want to use your platform? In this talk we'll discuss how we built Unicron, our Kubernetes-based MLOps platform that powers the machine learning use cases behind Adevinta's global online marketplaces. In particular we'll dive into the technology behind the platform, the teams that our powering it and the processes that we use to get regular feedback on new ideas and features. Additionally, we'll also demonstrate how we use golden paths coupled with Diataxis-flavored documentation to help users get their own use cases up-and-running within days, rather than months.