Liven is a database designed for applications that need to ingest, process, and store streaming data without the complexity of assembling multiple systems. It combines the durability of a traditional database, the query flexibility of a stream processor, and the semantic search capability of a vector database — all in one binary.
The project started from a simple observation: applications that handle real-time data often need streaming ingestion, query processing, durable storage, and semantic search — capabilities that typically require multiple systems. Liven eliminates that complexity by providing all four in a single query language with a single on-disk format.
Liven is built for the edge, for IoT devices, for AI agents, for real-time prediction markets — any environment where data moves fast and infrastructure needs to be minimal.
Built from the ground up for data that moves. Batch processing is a special case of streaming, not the other way around.
Embedded library, network server, TUI shell — same binary. No orchestration, no sidecars, no runtime dependencies.
Append-only storage with CRC32 integrity checks on every record. Crash-proof by design, not by ceremony.
Apache 2.0 licensed — free for any use, including commercial and proprietary embedding. No copyleft restrictions.
Optimized for the edge, IoT, and embedded use cases. ~1.5 MB core binary, microsecond point lookups, deterministic resource limits.
Liven is open source under the Apache 2.0 license. Contributions, issues, and discussions are welcome on GitHub.