JasonHonKL

JasonHonKL / PardusDB

Public

SQLite-like embedded vector database

pardusai.org
15
3
100% credibility
Found Feb 17, 2026 at 15 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

PardusDB is a single-file embedded database for storing vectors with metadata and performing fast approximate nearest neighbor searches using SQL-like queries.

How It Works

1
🔍 Discover PardusDB

You hear about PardusDB, a simple way to store and find similar items like document meanings in one easy file, perfect for local AI projects.

2
📥 Get it ready

Follow the quick setup guide to have PardusDB working on your computer in moments.

3
📁 Make your storage file

Create a single file to hold all your data, just like a supercharged notebook.

4
📋 Set up your list

Define a spot in your file for vectors (number patterns representing meanings) and extra details like titles or notes.

5
✨ Add your info

Put in your first batch of documents with their meaning patterns, and watch it organize everything smoothly.

6
🔎 Find matches

Ask for items similar to a new pattern, and get the closest matches ranked by how alike they are.

🎉 Fast private search

Enjoy lightning-quick searches on your own computer with full privacy, powering your AI apps effortlessly.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 15 to 15 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is PardusDB?

PardusDB is a Rust-built, SQLite-like embedded vector database that stores embeddings and metadata in a single file for local AI apps. Developers get familiar SQL syntax for creating tables with VECTOR columns, inserting data, running similarity searches via `embedding SIMILARITY [query_vec]`, plus JOINs, GROUP BY aggregates, and transactions—all without servers or external deps. Use the CLI REPL (`pardusdb mydb.pardus`), Python/TS SDKs, or MCP server for AI agents.

Why is it gaining traction?

It crushes benchmarks: 1983x faster inserts than Neo4j, 200x over HelixDB on 10K vectors, with single-search queries at 3µs on Apple Silicon. The hook is SQLite familiarity meets HNSW vector search in a zero-config binary, plus optional GPU batching and thread-safe concurrency for real apps. Pure Rust means tiny footprint and MIT license for commercial use.

Who should use this?

Backend devs building RAG pipelines or semantic search in local LLMs, where you need persistent vectors without cloud costs. Edge AI engineers on laptops/devices wanting SQL over embeddings for rec systems or knowledge bases. Prototype hackers testing vector workflows before scaling to LanceDB or PGVector.

Verdict

Try it for proofs-of-concept—blazing speed and SQL make it addictive, but 15 stars and 1.0% credibility signal early days with basic docs and unproven scale. Solid for small-to-medium datasets; watch for production hardening.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.