Metaflow

Netflix built Metaflow to let data scientists take ML and AI projects from notebook prototype to production without handing them off to another team, and the framework is now maintained with backing from Outerbounds. It presents a plain Python API for structuring workflows as versioned, resumable steps, tracking experiments and artifacts automatically, and scaling out to cloud compute on CPUs and GPUs when a laptop is no longer enough. One-click deployment pushes the same code to production orchestrators, and reactive orchestration lets workflows trigger on external events. The project reports use across thousands of projects at companies including Amazon, Doordash, Goldman Sachs, and Ramp, on top of more than 3,000 projects internally at Netflix. It installs from pip or conda-forge and is open source under the Apache 2.0 license. Data scientists and ML engineers who want infrastructure concerns abstracted behind ordinary Python are the core audience.