Building an ML Platform from Scratch
summary
A minimal ML platform is built end to end: a cookiecutter model template generates a FastAPI model server and training scaffolding, experiments and pickled models land in MLflow storing model artifacts to S3, training datasets are tracked with DVC for versioning, and the whole infra is declared with Pulumi+EKS infra-as-code. GitHub Actions builds and pushes Docker images to ECR and deploys to Kubernetes, ML servers serve predictions and MLflow + Postgres track metadata so rollbacks and audits work.
AI-written summary · about
chapters (9)
- 0:00Introduction and setup
- 1:39ML infrastructure needs
- 4:12Kubernetes explained
- 8:21Architecture overview
- 14:46Production flow details
- 15:50Infrastructure as code
- 19:39Key workshop concepts
- 20:01Code demonstration
- 28:52Resources and closing
related talks
32mTeam Teardown: Evolving an ML Platform
37mPlatform for Complete Machine Learning Lifecycle
28mBuild an End-to-End ML Workflow for Your Organization
43mMachine Learning Engineering principles with Python and MLFlow
41mFrom Prototype to Production Building a Scalable MLOps Platform from a Reference Arch
31mMLflow: Platform for Complete Machine Learning Lifecycle
24mBuilding an ML Application Platform from the Ground Up
01mE2E ML Platform on Kubernetes with just a few clicks