Sean is a tech advocate for Preferred Networks, a Berkeley-based robotics company. He teaches python and machine learning, writes tutorials, and connects with the greater developer community in the Bay Area. In his spare time, he enjoys volunteering at PyLadies, hacking on open source, and traveling.
This workshop will have four or so Lego Mindstorms that we'll use to learn how to program a Mindstorm to use its camera to follow a racetrack made out of tape.
Using a real-time OS, we will learn about computer vision using the tools made by Lego. Time permitting, we will then remove parts of the racetrack and use machine learning to improve our race car.
Learn how the open source Python module CuPy works as a drop-in replacement for NumPy to enable calculation using GPUs. We'll explain how engineers can speed up calculations by making their own kernels on the GPU. We'll also cover other projects incorporating CuPy to increase calculation speed.
CuPy is not just a wrapper for CUDA. It is designed to be a drop-in replacement for NumPy to make maintaining your own library even easier. This talk will explore the origins of CuPy and why we needed more than just PyCUDA for Chainer.