awesome production machine learning

The list covers the full production ML lifecycle: model serving and deployment, experiment tracking, feature stores, data versioning, model monitoring and drift detection, testing and validation, CI/CD for ML, and explainability.

The collection is organized by operational concern rather than model type, reflecting the reality that production ML challenges are largely infrastructure and process concerns rather than algorithmic ones.

Categories include orchestration tools (Airflow, Prefect, Kubeflow), serving frameworks (TorchServe, Triton, BentoML), monitoring solutions (Evidently, WhyLogs, NannyML), feature platforms (Feast, Tecton, Hopsworks), and data validation libraries (Great Expectations, Pandera)with brief descriptions that help practitioners distinguish between overlapping tools.

ML engineers building production systems who need to survey the tooling landscape before making technology selections, engineering managers designing MLOps infrastructure for growing data science teams, and practitioners new to production ML who need to understand what categories of tooling exist and why use this list as an orientation resource.

The active GitHub maintenance means new tools are added as they gain community adoption, and the curation quality has made it one of the more trusted aggregations of MLOps tooling relative to the many lower-quality 'awesome' lists in the space.