Write You a Vector Database is a short, Rust-first systems course. Build a small in-memory vector database in Rust, compare approximate results with exact search, and connect the resulting indexes to SQL through DataFusion.
The course focuses on the boundary where algorithms become database features:
in-memory table → DataFusion optimizer rule → IVFFlat → NSW → HNSW → IVF-PQ → benchmark
Instead of hiding vector search behind an HTTP API or an ANN library, the course exposes the algorithms, evaluation contracts, query planning, and execution boundary that make SQL vector search work.
The Rust course has six required chapters: an Arrow-backed in-memory table and safe DataFusion optimizer rule, IVFFlat, NSW, HNSW, residual IVF-PQ, and a final shared recall and latency benchmark. The implementation chapters include starter code, focused tests, and separate completed reference crates. Chapters 1–4 include SQLLogicTests; Chapter 5 includes a focused SQL planner/EXPLAIN test.
The final benchmark compares Flat, IVFFlat, NSW, HNSW, and IVF-PQ on one shared Euclidean workload. It reports build time, recall, p50 and p99 search latency, plus the IVF-PQ search-representation accounting.
Run the completed reference with:
cd rust
cargo test -p vector-core -p vector-datafusion
cargo run --release -p vector-core --example recallThe original C++/BusTub edition is deprecated and unmaintained. It remains online for existing readers but is no longer recommended for new learners.
Join skyzh's Discord server to study with the write-you-a-vector-db community.
The code in this repository is licensed under the Apache License 2.0. The book, including its Markdown and figures, is licensed under CC BY-NC-SA 4.0.
The bustub-vectordb-starter and bustub-vectordb-solution git submodules retain their own upstream license terms.