fast.ai

fast.ai was founded in 2016 by Jeremy Howard and Rachel Thomas to make deep learning more accessible to practitioners and researchers outside elite academic settings. The institute publishes free "Practical Deep Learning for Coders" and "From Deep Learning Foundations to Stable Diffusion" courses, taught annually and watched by hundreds of thousands of learners.

The fastai library is a PyTorch wrapper providing high-level APIs for vision, text, tabular, and collaborative-filtering models, used in production and in academic research. The associated nbdev tool builds Python libraries from Jupyter notebooks.

fast.ai operates as a small non-profit team led by Jeremy Howard, with research output published as papers, library releases, and open course material rather than commercial products.