AutoKeras is an open-source AutoML library built on Keras. It searches neural network architectures and hyperparameters automatically, with task-level entry points for image classification and regression, text classification and regression, structured data, and multi-modal problems.
The appeal is the shape of the API. Rather than a platform to adopt, it is a few lines of Python that turn a supervised learning problem into a fitted model without hand-designing an architecture – which makes it genuinely useful as a strong baseline. Knowing what an automated search achieves on your data tells you whether a hand-built model is worth the effort, and often it is not.
It is Apache-2.0 licensed with around 9,300 GitHub stars and full tutorials. Check the activity before depending on it: the repository is not archived, but as of this review the most recent commit was November 2025 – stable and usable rather than actively developed. Being a library, it also assumes Python and a working data pipeline; there is no interface and no support contract.
