GitLake: Git-for-data for the agentic lakehouse

Weiming Sheng; Jinlang Wang; Manuel Barros; Aldrin Montana; Jacopo Tagliabue; Luca Bigon

GitLake: Git-for-data for the agentic lakehouse

数据库 2026-07-09 v1 人工智能

作者: Weiming Sheng , Jinlang Wang , Manuel Barros , Aldrin Montana , Jacopo Tagliabue , Luca Bigon

摘要

We present GitLake, a Git-for-data design for an agent-first lakehouse. The system lifts single-table Iceberg snapshots into lakehouse-wide commits, branches, and merges, letting agents work on isolated branches while humans review and publish changes. Pipelines run on temporary branches and publish through a final merge, so all outputs become visible atomically or none do. Finally, we report production lessons as well as correctness insights from a preliminary Alloy model of our core abstractions.

引用

@article{arxiv.2607.08319,
  title  = {GitLake: Git-for-data for the agentic lakehouse},
  author = {Weiming Sheng and Jinlang Wang and Manuel Barros and Aldrin Montana and Jacopo Tagliabue and Luca Bigon},
  journal= {arXiv preprint arXiv:2607.08319},
  year   = {2026}
}

备注

Pre-print of the paper accepted at DASHSys, VLDB 2026, Boston, USA