Stratum – SQL that branches and beats DuckDB on 35/46 1T benchmarks
Git-like branching for columnar data with DuckDB-beating benchmarks from pure JVM.

Java Vector API beats DuckDB on 35/46 queries; O(1) table branching via structural sharing.
Java-based data platforms, analytics engineers, Datahike ecosystem users needing SIMD-optimized SQL
DuckDB · Presto/Trino · Polars
Git-like branching for columnar data with DuckDB-beating benchmarks from pure JVM.
Replaces PostgreSQL + Redis + Neo4j + ClickHouse with one Rust binary.
Using a single-file .pardus format with CREATE/INSERT/SELECT + SIMILARITY queries gives a very familiar developer UX for embedding storage. The combination of graph-based ANN, full transactions, thread-safety, and zero external dependencies is an uncommon and useful engineering combo for local-first AI work; it would win more attention with benchmark comparisons and richer ecosystem integrations (connectors/clients).
Zero-install Postgres diagnostics, but pg_stat_statements and pgAdmin already exist.
MegaHAL chatbot runs in pure SQL—no PL/pgSQL escape hatches, just CTEs and recursive queries.
Graph RAG without Neo4j — pure vector search beats HippoRAG on multi-hop benchmarks.