MiniQdrant Tutorial

English quick start · 中文版

MiniQdrant is a direct-first Python reference implementation of filtered vector search, versioned immutable segments, online optimization, and durable recovery. It exposes the mechanisms behind a single-node vector database without reproducing Qdrant's network API or distributed deployment surface.

中文简述：MiniQdrant 用 Python 展示过滤向量搜索、版本化不可变分段、在线优化 与持久恢复；它聚焦单节点机制，不复刻 Qdrant 的网络与分布式产品表面。

Install

git clone https://github.com/system-in-miniature/MiniQdrant.git
cd MiniQdrant
uv sync

Python 3.12+ and uv are required.

First experiment

uv run python -m miniqdrant.labs.filtering

The lab searches a tiny collection with a tenant filter. It prints matching IDs [1, 3], excluding a vector-similar point from another tenant, and exposes the selected per-segment plan.

Continue with the architecture tour, then compare modules with Qdrant.

For the complete API, CLI, scope, and reliability boundary, read the repository README.