Dagster is an open-source data orchestrator built around software-defined assets: instead of only wiring tasks, you declare the data assets (tables, files, models) your pipelines produce and the dependencies between them. Scheduling, materialization, and lineage follow from that graph.
The framework emphasizes engineering workflow — local development, unit-testable pipelines, typed configuration — and integrates closely with dbt, exposing dbt models as assets alongside ingestion and ML steps.
Dagster suits teams that want an asset catalog and observability built into orchestration rather than bolted on. It can be self-hosted, with Dagster+ as the managed commercial offering; it is most commonly evaluated against Airflow and Prefect.