BACK TO VIDEOS
Next VIDEO
Next VIDEO
Git for Data: How Agents Write to Production Without Breaking It
Agents can write to production data fast, in parallel, and sometimes wrong. Git for data brings the software workflow, branch, review, merge, and roll back, to tables in object storage, so an agent works on an isolated branch that production never sees until a human approves the merge.
Chapters
0:00 - Introduction and what Git for data is
1:49 - Why agents change the problem: access, control, and table sprawl
6:42 - Under the hood and fitting your stack: branches, Iceberg, Databricks
15:22 - Live demo: a Slack request from ingestion to dashboard
33:24 - Review and merge: the PR, audit, and the data team's new role