From Side Project to Startup: Graph Lessons From the Cartography Journey
summary
Cartography starts as a Python service that pipes cloud provider data into Neo4j + Python and evolves into an OSS project and YC-backed startup; the team implements UNWIND batching plus managed transactions and explicit consume calls to avoid autocommit failures and all-node scans. The graph model encodes CVE-to-image lineage and team ownership, drives automated ticketing and reconciliation, and wins an $800M-class procurement outcome, though scaling forces a custom ORM/codegen and still-ongoing refactors.
AI-written summary · about
chapters (14)
- 0:00Introduction and speaker background
- 2:32From red team to open source
- 5:20Building Cartography at Lyft
- 7:07Business needs vs security tools
- 9:12Neo4j challenges and reliability
- 10:57Killer app at Lyft: vulnerability management
- 15:30Building engineering scaffolding
- 16:14Neo4j performance optimizations
- 19:10Custom ORM for Neo4j
- 21:59Lessons from open source to startup
- 23:45Transitioning to startup life
- 25:11Founding a security startup
- 25:52Future directions and graph RAG
- 27:18Closing takeaways and Q&A
