Kaos Simplifies MLOps Model Deployment Across Kubernetes

Open-source project kaos, developed by Alejandro Saucedo, offers a Heroku-like CLI to package, train, and serve ML models without writing Kubernetes configuration. The project, active primarily in 2021–2022, enforces infrastructure-as-code for reproducibility and aims to reduce the industry’s deployment gap—VentureBeat estimates 87% of data science projects never reach production. Kaos targets smaller teams seeking simpler MLOps on Kubernetes.