FSDL Bootcamp — Introduction

A quick introduction of me: I am Gerard Simons, I live in Rotterdam where I co-founded a company called Captain AI. We train an AI to pilot boats better than any human can. This involves a lot of machine learning and especially deep reinforcement learning.

To up my skills I attended my first ever bootcamp this summer at UC Berekeley. It was a pretty great experience, learning so much it made my head throb (that’s good!), meeting a whole bunch of like-minded people, and generally just having a very good time.

It all started with a tweet from a legend of the field, Pieter Abbeel:

This got me and my colleagues very excited. Looking at the speakers, the content and especially its focus towards “full stack” intrigued us. This looked like it could be a very interesting course.

Different than other courses, this one aimed to focus more on the practical aspects of tackling a machine learning challenge. Questions to be answered during this course would be : How do you structure your project, what useful best practices are there and how do you troubleshoot any issues that may arise. Most other courses are much more focused on the math / CS background or the training part. The contents of this course can be thought of as another layer on top of the existing material, as shown by the figure below.

How the bootcamp’s course relates to existing material

For those who could not attend I hope this blog post will give you a bit of an insight what it was like to be there and pass on some of the knowledge gained.

Each day has it’s own little blog post:

Here is a schedule of the entire bootcamp so you can figure out which day / blog post you might wanna check out first.

Bootcamp schedule: Pretty packed, but pretty good