I’m Matt. I’m a programmer, writer, and entrepreneur. I’ve worked on open-source at Google, and as a developer at a private equity firm in New York.
I have a BA in Mathematics from Columbia University, and working towards an MBA at Stanford Graduate School of Business (2021).
- Google Cloud at KubeCon 2018: Interview with Matt Rickard
- The Kubernetes Podcast by Google: Episode #6
- Live Demos from KubeCon: Matt Rickard
- Open Source Summit: Alternative Frontends with Buildkit, DockerCon 2019
- Kubeflow Fairing, Kubeflow Contributor Summit 2019
- Building Docker Images Without Docker, KubeCon EU, Copenhagen 2018
- Developer Tools, Kubernetes Developer Summit, Copenhagen 2018
- Kubernetes Development Workflows, CloudCamp SF 2018
- Easy Kubernetes Development with Skaffold and Redis, RedisConf 2018
- Skaffold, Kubernetes Bay Area Meetup 2018
- Kubernetes Developer Tooling, DevOps Night SF 2018
- Container Tools Overview, Google Developer Expert Conference, 2018
- Minikube Developer Workflow and Advanced Tips, KubeCon Austin 2017
- Kubernetes 1.4 Release Meetup - Minikube, CoreOS 2016
Minikube is the developer tool for Kubernetes. It runs an entire single-node distributed system on any laptop. I contributed a lot f major features to the project, building out the actual provisioning strategies and using native hypervisor frameworks to boot a custom, lightweight linux distribution that ran Kubernetes.
Skaffold was created from my frustration developing applications that run on Kubernetes. It’s a cloud-native application development tool focused on speeding up the inner developer loop. File syncing to Kubernetes deployments, automatic cached docker builds, and smart deployments make application development quick and repeatable on any machine. It supports local, cloud, and hybrid development.
Kubeflow is a machine learning toolkit for Kubernetes. It allows developers and operators to build out distributed training and inference machine learning jobs on Kubernetes infrastructure. I focused on the experience of going from a Jupyter notebook to a distributed training job automagically.
An alternative to the Dockerfile that can be built with the existing Docker CLI. It uses a proof-of-concept YAML declaration for images and works through the native buildkit API found in every Docker distribution.
A set of rules and minimal docker images that can be used for language runtimes that don’t contain all of the extra things found in a Linux distribution. Instead of starting from Ubuntu, we build only the necessary files into a container image tarball. It also can be used to build docker images on unprivileged infrastructure, like building docker-in-docker without any extra permissions.