Knowledge Graph Insights

Alexandre Bertails: The Netflix Unified Data Architecture – Episode 40

Nov 3, 2025
Alexandre Bertails, a software engineer at Netflix's content engineering team, talks about pioneering the Unified Data Architecture (UDA) to enhance semantic interoperability through RDF. He explores the challenge of creating a singular schema for diverse internal needs while detailing the innovative Upper domain modeling language. Bertails highlights Upper's self-describing and self-governing traits, its compact design, and how Netflix operationalized these concepts to improve data management and accessibility across teams. Tune in for insights on making complex data relatable!
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

Model Once, Represent Everywhere

  • Netflix designed UDA to capture one domain model and project consistent representations across GraphQL, Avro, Java, and SQL.
  • This enables semantic interoperability so the same concept can be used across many internal data formats.
INSIGHT

Mappings Are First-Class Graph Objects

  • UDA's knowledge graph contains domain models, mappings, and data container metadata like table coordinates and schema.
  • Mappings connect conceptual types to concrete data sources and API resolvers for precise interoperability.
INSIGHT

Conceptual RDF And Virtual Knowledge Graphs

  • UDA centers on conceptual RDF but treats instance data virtually rather than fully materializing it.
  • They use RDF's conceptual framework to reason about schemas and project them into many representations.
Get the Snipd Podcast app to discover more snips from this episode
Get the app