Embeddable, in-memory, document-oriented database with a high-level Query builder interface.

VectorRAG.Net is a .NET-native high-performance vector database library for semantic search and RAG (Retrieval-Augmented Generation). Core search is based on Random Hyperplane LSH candidate generation with exact rerank by dot/cosine.

Near-optimal vector quantization from Google's ICLR 2026 paper — 95% recall, 5x compression, zero preprocessing, pure Python FAISS replacement

Multi-modal embeded database for edge devices

Code and results for "Revisiting RaBitQ and TurboQuant: a symmetric comparison of methods, theory, and experiments".

A (not very) frequently updated list of ANN vector search papers on declarative recall through early termination, published in top data management venues.

Production-ready multimodal retrieval system built with OpenCLIP, Qdrant, FastAPI and Streamlit. Includes full evaluation pipeline (Recall@K, mAP, nDCG) and Docker-based deployment.

Approximate shortest path between the nodes in a (knowledge or any other) graph

Build-time, multi-agent RAG pipeline that turns raw course materials into structured topic pages, prerequisite graphs, and QA reports

A framework for the implementation of candidate generation/retrieval algorithms for recommender systems.

RAG pipeline prototype MVP project

A basic RAG pipeline which uses gpt-oss-20b model to answer the user query with the external knowledge stored in a vector database.

🔥 VectorForge - Lightweight High-Performance Vector Index Engine | Pure Python, Zero Dependencies, HNSW/IVF/LSH/BruteForce, TUI Dashboard, Multi-Format I/O, Cross-Platform

Building a Custom Vector Search Engine with Weaviate : The project discusses the architecture of Weaviate, an open-source vector database and provides a tutorial implementation of a custom vector search engine using Weaviate Cloud Service(WCS).

Comprehensive vector database benchmark: pgvector, Pinecone, Weaviate, Chroma, Qdrant, Milvus — recall@k, QPS, latency, cost and scalability comparison at 100M vectors

An advanced implementation of billion-scale Approximate Nearest Neighbor (ANN) search. Features include a modular multi-quantizer pipeline (PQ/OPQ/AVQ), shared-memory safe Cuckoo filtering, and speculative 2-Hop CPU prefetching for zero-latency graph traversal

A WebGPU-powered vector database for local semantic search, exact similarity queries, and benchmarked embedding workflows.

Transform-domain representation enabling 3–4× storage reduction with direct ANN search and novel multi-resolution signals. UK patent application under accelerated examination (Green Channel).