Under the Hood
I developed content for Cruise’s 2021 Under the Hood event. See the AI team’s presentation below:
I was on the original hackathon team which prototyped webviz, Cruise’s browser-based visualization tool. Since then, it has become a ubiquitous tool within the company. Former colleagues have since gone on to found Foxglove, which is a fork of Webviz aimed at enterprise support for robotics visualization.
I ended up being the person who put together the final puzzle pieces to compile TensorFlow for the Raspberry Pi back in 2015. While TensorFlow has officially supported binaries for Raspbian these days, at the time I was personally putting out unofficial releases for people to download. Bazel’s cross-compiling capabilities were a nightmare to figure out, which meant that compiling TensorFlow from scratch on a Raspberry Pi was the best way to handle it. The main problem is that it took over eight hours to compile! And you needed to install a USB drive as swap space.
Hopefully you won’t blame me when I say that I’m probably happier now that I don’t have to manage this. That said, I do get some nostalgia for that time, and I still have colleagues and contacts that come up to me who find my repository when they start experimenting with machine learning on their RPi.
I cannot take all (or even most) of the credit for this effort. This links to the credits from the original README which recounts the history of how we cracked that nut..
I was an early contributor to TensorFlow after it was open sourced in late 2015. I’m named in the following releases:
Prior to getting involved in the TensorFlow community, I was a freelance artist for Illumination, primarily focusing on 2D animation for digital media and animatics.