ethicalml/awesome-production-machine-learning

This is a community-maintained directory of open source libraries that help teams take machine learning out of a researcher's notebook and into a real product. The focus is the part of the lifecycle that comes after a model first works: deploying it, monitoring it in production, versioning it, scaling it, and keeping it secure. The project is laid out as one long page grouped into roughly two dozen categories so you can jump straight to what you need. Categories include AutoML, data pipelines and stream processing, deployment and serving, evaluation and monitoring, explainability and fairness, feature stores, model training and orchestration, metadata and experiment management, model storage optimisation, privacy and safety, and industry-strength sections for computer vision, natural language processing, recommender systems, reinforcement learning, anomaly detection, information retrieval, robotics, and visualisation. Each entry is a short blurb pointing at the library's own GitHub repo, with a live star-count badge. As a concrete example, the AutoML section names projects such as AIDE, AutoGluon, AutoKeras, auto-sklearn, Ax, BoTorch, EvalML, Feature Engine, Featuretools, FLAML, HEBO, Katib, and keras-tuner, each with a one-line note on what it does. People reach for this list when they are evaluating tooling for an MLOps stack, comparing options inside a single category, or simply trying to learn what serious production machine learning looks like outside the model-training step. A separate Awesome Production GenAI list covers generative AI tooling, and a search toolkit hosted on Hugging Face Spaces helps filter the catalogue. A ten-minute video overview, a newer 2024 state-of-MLOps video, and a Machine Learning Engineer newsletter round out the resources.