Kùzu: A Graph Database Management System for Python Graph Data Science

on YouTube ↗31m

" summary

Kùzu is an open‑source, disk‑backed graph DBMS built as an embedded library for Python graph data science that installs into notebooks and scripts. It ingests raw files and pandas frames, runs Cypher queries, and exports with no‑copy to NetworkX and PyTorch Geometric and Arrow/pandas. The system claims a query execution trick that compresses intermediate results from exponential to linear growth on certain pattern joins, provides a web explorer, and ships under an MIT license.

AI-written summary · about

" chapters (12)

1. 0:00Kùzu introduction
2. 3:40Graphs vs tables for modeling
3. 6:32Kùzu embedded system
4. 9:02Kùzu fast pattern architecture
5. 11:53Kùzu with Python tools
6. 12:37Demo: Basic operations
7. 17:16Demo: NetworkX querying
8. 20:24Demo: Community detection
9. 24:27Comparison with commercial graph databases
10. 25:59AI integration roadmap
11. 27:29Demo: Web Explorer
12. 31:00Closing & Q&A

" related talks

• 49mModern Graph DatabasesDavid Ostrovsky · Build Stuff 2025
• 41mGraphGen: Conducting Graph Analytics over Relational DatabasesKonstantinos Xirogiannopoulos · PyData DC 2016
• 46mGraph Databases: Talking about your Data Relationships with PythonNicolle Cysneiros · PyCon 2018
• 29mImplementing a Graph Database based on PostSQLKisung Kim · PGConf Silicon Valley 2016
• 26mGraph Databases will change your (freakin') lifeElena Williams · PyConline AU 2020
• 30mGraph Data SciencePaco Nathan · PyConline AU 2021
• 01h 00mThe Open-Source Distributed Graph Database: Nebula GraphWey Gu · Percona Community Live 2022
• 49mGraph the planet: Wrangling GPU graph dataframes with GFQLSindre Breda · NDC Oslo 2025