Speaker : Anta Aidara Machine learning is no longer just for snake charmers (pythonista) or those with access to powerful computational resources. As part of the PAIR project, the Google Brain team released last year the deeplearn.js library which has since evolved into TensorFlow.js. Tensorflow.js takes advantage of WebGL to allow among other things to build and train neural networks in the browser from scratch or using techniques such as transfer learning. Also a recent release of Tensorflow backend for Tensorflow.js now allows you to fully take advantage of the hardware (GPU, TPU, CPU ) directly on the server side via Node.js. I will explain you during this talk, based on my own experience, how I took my progressive web app to the next level using machine learning as well as the possibilities available to implement machine learning into your browser.