Glossary

Key terms and technologies referenced throughout this documentation.

Term	Definition
Medallion Architecture	A layered data-quality pattern (Bronze → Silver → Gold) within a lakehouse, ensuring progressive refinement of data.
Apache Iceberg	An open table format for large analytic datasets providing ACID transactions, schema evolution, and time-travel.
Apache Kafka	A distributed event-streaming platform used for real-time ingestion and Change Data Capture (CDC).
Apache Spark	A distributed in-memory processing engine for large-scale ETL and ML workloads.
dbt	A SQL-first transformation framework enabling modular, tested, version-controlled data models.
Trino / MPP OLAP	A massively-parallel query engine for fast interactive analytics and federation across sources.
DataNature Atlas	The governance, security, and semantic layer that wraps the platform — catalog, RBAC, lineage, PII, and audit.
DataNature Voir	The multi-agent AI platform — Chat, Twins, Skills, and Connectors — for business and platform builders.
DataNature AI360	The decision-intelligence component — Customer 360, segmentation, semantic layer, and reverse ETL.
DataNature Marketplace	A catalog where partners and ISVs publish models/AI solutions that run governed on-platform and become Voir agents.
Semantic Layer	A governed abstraction exposing consistent KPI definitions and business views to all consumers.
Reverse ETL	Pushing enriched data from the platform back into operational tools (CRM, marketing, etc.).
CDC	Change Data Capture — capturing row-level INSERT/UPDATE/DELETE changes from source databases in near-real time.
RAG	Retrieval-Augmented Generation — grounding model responses with retrieved documents from a knowledge base.
RBAC	Role-Based Access Control — least-privilege access policies applied per user role across all layers.
Vector Store / KB	A database of embeddings enabling semantic similarity search and grounding AI agents with document knowledge.