LinkingMem — Graph-native RAG Engine
LinkingMem — Graph-native RAG Engine
Start new thread
Kent Phung

LinkingMem — Graph-native RAG Engine - LinkingMem — Graph-native RAG Engine

by
LinkingMem — Graph-native RAG Engine
•
LinkingMem is a Graph-native RAG engine combining Rust performance with Python AI plugins. It unifies vector search (HNSW), graph traversal (BFS), and LLM reasoning in a single pipeline for fast multi-hop retrieval. Key differentiators include tight graph+vector integration, embedding-based entity resolution, pluggable LLM/embedding backends, mmap-based low-latency storage, and production-ready scalability for large knowledge graphs.

Add a comment

Replies

Best
Kent Phung
Maker
📌
I built LinkingMem because I kept running into the same issue when working with RAG systems: vector search alone is not enough when the question needs relationships or multi-step reasoning. Most systems can find similar text, but they struggle when you ask things like “who works with X?” or “how are A and B connected?”. I wanted something that could actually follow connections, not just match embeddings. At first I only used a simple vector database, but it quickly became clear that it couldn’t handle relationship-based queries well. So I added a graph layer on top and made the system combine both vector search and graph traversal in one pipeline. The final design is pretty straightforward: embed the query, find nearest nodes, expand through the graph, rank results, then send the best context to the LLM. Rust is used for performance-critical parts, and Python handles the AI models. While building it, I kept simplifying the system instead of making it more complex, until it could run fast, handle large graphs, and still give better answers for multi-hop questions.