Video URL: https://www.youtube.com/watch?v=J4zvLJn3ya8
00:01:22 Introduction
00:02:03 Exax history: From startup to eBay to open source
00:02:55 Two design goals: Minimise risk for manual mistakes and maximise efficiency
00:03:34 The Backblaze dataset example
00:04:23 Backblaze dataset: Introduction to animation
00:05:37 Backblaze dataset: Animation starts
00:06:27 A program execution
00:06:58 Collect everything
00:07:38 Reuse (Only compute when seeing something new)
00:08:33 Input parameters can be links to existing jobs
00:09:17 Compare traditional tools to Exax
00:10:16 Jobs reflect different stages of source code
00:11:01 A key idea
00:11:38 Introduction to transparency and reproducibility
00:11:47 Transparency
00:12:22 Transparency examples
00:12:45 Reproducibility
00:13:56 Backblaze dataset: Back to example
00:14:33 Backblaze dataset: Intermediate files
00:15:56 Backblaze dataset: Execution times
00:17:53 NYC Yellow Taxi trip dataset example
00:18:18 Taxi dataset: Quality
00:18:50 Taxi dataset: January 2016 US blizzard
00:19:09 Taxi dataset: Introduction to animation
00:19:20 Taxi dataset: Animation starts (COVID impact on taxi business)
00:20:07 Taxi dataset: Execution times
00:20:46 Other features
00:21:02 High-performance data access
00:23:13 Parallel processing and data locality
00:24:50 Conclusion
00:25:01 Links to GitHub and resources