Alice in the world of machine learning

on YouTube ↗41m

" summary

Builds a Kubeflow proof of concept on Google Cloud that containers three components and runs them via Argo-backed pipelines: Kubeflow on GCP, Dockerized components, preprocess -> train -> deploy pipeline. Uses YAML component specs, a Python DSL compiler to emit pipeline manifests, pushes images to GCR, and stores artifacts in GCS/persistent volumes. Runs show initial failures then improved accuracy; a live test classifies “weather in London today” into city/time/entity correctly. Describes serving options (Seldon/TensorFlow Serving) and where operators and notebook services map to Kubernetes.

AI-written summary · about

" chapters (9)

1. 0:00Start
2. 2:57Immutability
3. 3:25Declarative Nature
4. 11:24Ml Flow
5. 16:25Kubeflow Pipelines Ar
6. 18:17Tensorflow Jobs
7. 25:25Docker File for Building a Docker Image the Bash Script
8. 29:51Compiling the Pipeline
9. 30:15Create an Experiment

" related talks

• 46mDeploying scalable Machine Learning models with KubernetesAlessandro Vozza · Future Tech 2019
• 54mScaling Machine Learning Pipelines in CloudSalman Iqbal · NDC Oslo 2021
• 01h 26mTutorial: Introduction to Kubeflow PipelinesMichelle Casbon, Dan Sanche, Dan Anghel · KubeCon + CloudNativeCon Europe 2019
• 36mMachine learning pipeline Kubeflow, let's startGabriele Santomaggio · containerday 2020
• 51mMachine Learning using Kubeflow and KubernetesArun Gupta · Devoxx Belgium 2019
• 22mBuilding a MLOps Platform with Kubeflow on GKEJoinal Ahmed, Nikhil Rana · Kubernetes Community Days Bengaluru 2023
• 31mKubeflow - Data Science on SteroidsSascha Grunert · ContainerDay 2019
• 54mEconomics and Best Practices of Running AI/ML Workloads on KubernetesMaulin Patel · KubeCon + CloudNativeCon Europe 2019