Have you ever felt like it’s getting increasingly more difficult to become an expert in new technologies? Even if you are only looking to quickly try the hello-world example on your laptop without giving away credit card information. Projects currently labelled as “open-source”, with hundreds of thousands of community members, are somehow inaccessible and distant. At the same time, the source code is available to anyone, so what’s the problem?
Cloud-native technology, specifically software designed to run in distributed systems, is very challenging to manage. While upstream communities can provide packaged solutions and comprehensive installation manuals, the matrix of configuration options is overwhelming! Who knows how to turn all the knobs and press all the buttons in a Kubernetes cluster to make it run optimally, and how much does it cost to learn how to do so?
This talk explores how we can learn from the path that Linux took to go from an inaccessible system, limited to a few in the 90s, to being in everybody’s pockets nowadays. It proposes a framework that upstream communities, especially ML tooling, can utilize to make their software more accessible to people who want to use the technologies without being drowned in a sea of YAML. It focuses on Kubeflow, a machine learning operations (MLOps) platform that is designed to run AI at scale, but it can be extended to other cloud-native applications such as Spark, Kafka or MySQL.
Join the community
---
Discord: / discord
GitHub: github.com/osc...
Open Collective: opencollective...
All links and social media: linktr.ee/osca...
Негізгі бет Democratizing Modern Open Source: Lessons from the ML World - Andreea Munteanu, Pedro Cruz
Пікірлер