Put open source to work
July 16–17, 2018: Training & Tutorials
July 18–19, 2018: Conference
Portland, OR

Live-coding madness: Let's build a deep learning library.

Joel Grus (Allen Institute for Artificial Intelligence)
2:35pm3:15pm Wednesday, July 18, 2018
Live coding
Location: Portland 252
Level: Intermediate
Average rating: ****.
(4.57, 7 ratings)

Who is this presentation for?

  • Data scientists, software engineers, and machine learning engineers

Prerequisite knowledge

  • A working knowledge of Python
  • A basic understanding of machine learning

What you'll learn

  • Understand deep learning concepts and abstractions
  • Learn how to design a library and how to use mypy to develop faster and safer


Joel Grus live-codes a deep learning library from scratch—well, from NumPy—and trains some demonstration models, placing particular emphasis on writing readable code, creating a usable library, and using good abstractions. You’ll learn a good bit about both deep learning and library design.

Joel explores tensors, loss functions, neural nets, and their layers, optimizers, and training methods. He’ll try to convince you of the virtues of Python 3, using type hints, and running a static type checker. You’ll also learn Joel’s own Python workflow, which will give you some ideas about how to improve yours.

Join in to discover that deep learning isn’t actually all that hard as long as you approach it thoughtfully.

Photo of Joel Grus

Joel Grus

Allen Institute for Artificial Intelligence

Joel Grus is a research engineer at the Allen Institute for Artificial Intelligence and the author of the beloved O’Reilly book Data Science from Scratch and the blog post “Fizz Buzz in TensorFlow.” Previously, he was a software engineer at Google and a data scientist at a variety of startups. He lives in Seattle.