DataFusion
Apache project: a Rust-written query engine on top of Arrow — like DuckDB but as a library, not a database, and increasingly the engine inside other systems.
Full SQL + DataFrame API, vectorised execution, reads Parquet/CSV/JSON/Arrow. Designed as a building block: InfluxDB 3.0, GreptimeDB, Comet (a Spark accelerator) and Ballista (distributed compute) all sit on top of DataFusion.
Not a standalone tool like DuckDB — you embed it in Rust or Python code. A good fit when you need SQL over Arrow data inside your own application.