HelixDB From Zero

Stop stitching three databases together. HelixDB is a Rust-native graph plus vector engine that holds your nodes, your embeddings, your typed edges, and your key-value documents in one process — no separate Postgres, no separate Qdrant, no separate Neo4j. This course walks a Rust-fluent engineer from helix init through a typed HelixQL schema, the helix check and helix compile pre-deploy gates, side-by-side graph traversal and vector search in the same query language, and a typed Rust client that calls a live HelixDB instance with four runtime contracts. Every primitive you meet is wired into a working .hx schema and a running engine you can install in one command. Module 4 puts graph traversal verbs (Out, In, WHERE, ORDER, RANGE) and vector top-k search side by side in the same query language, and adds ShortestPathDijkstras with composable weight expressions for cost-aware routing. You finish with the canonical hybrid-RAG pattern — SearchV returns top-k embeddings, then a typed edge climbs back to the source documents — shipped as a typed Rust client (helix-rs plus serde) with four runtime assertion contracts. The course closes with an honest read on where HelixDB is the wrong choice: columnar OLAP, Spark execution semantics, multi-tenant analytical warehouses.