Launched this week
LoraDB
Embedded Rust graph database for AI, Cypher & Vectors Search
2 followers
Embedded Rust graph database for AI, Cypher & Vectors Search
2 followers
LoraDB is an embedded graph database for connected systems. Built in Rust, it runs in-process, speaks a Cypher-like query language, and now supports first-class vector values for graph-shaped AI retrieval. Store relationships, entities, and embeddings in one engine across Rust, Node.js, Python, WASM, Go, Ruby, or HTTP.
I started LoraDB because I kept reaching for a graph database in places where the existing options felt too heavy for the job.
The workloads were always similar: connected entities, typed relationships, short traversals, ranking and filtering over neighborhoods, and state that needed to stay close to the application instead of across a slow boundary. Sometimes it was product data, sometimes internal tools, sometimes agent memory. The model kept becoming a graph, but I did not want a large database stack before I even knew the model fit.
So I built LoraDB to feel different on day one: local, in-process, fast to start, small enough to understand, and queryable with a Cypher-like language.
LoraDB is an embedded graph database written in Rust. It runs in memory, speaks a pragmatic Cypher-like query language, and works across Rust, Node.js, Python, WASM, Go, Ruby, and HTTP.
The reason that matters is simple: a lot of AI retrieval systems need both similarity and structure. Similarity finds candidates, but the graph explains them. I wanted embeddings to live next to the relationships that give them meaning, not in a separate system glued together by application code.
If you try it, I’d love to hear what graph you’d load first, what kind of retrieval or relationship-heavy workflow you’d test, and most important; tell me what you think!