V

vector-database

PaveDB is an inspectable retrieval database for AI applications: text-backed vector search with source provenance and query replay, embedded or over HTTP.

Website: https://pavedb.org/

Persistent vector memory for AI agents with decay and priority, wrapped in MCP server.

Python SDK for PaveDB’s /v1 API over HTTP, local embedded, or ephemeral mode, covering collections, ingestion, search, query replay, chunks, embedders, and archives.

MCP server (streamable HTTP) for hybrid dense + SPLADE++ search over a Qdrant profile knowledge base

MindVault is an open modular knowledge infrastructure specification designed as the knowledge layer for autonomous AI systems. It provides a foundation for vector storage, semantic understanding, persistent memory, reasoning preservation, and intelligent retrieval, enabling AI agents and RAG systems to build, manage, and utilize long-term knowledge. https://roxanneardary.com/mindvault/

Curated vector databases, vector search engines, and embedding storage systems for AI.

Flexible GraphRAG: Python, LlamaIndex (LangChain also coming) Docker Compose: 8 Property Graph dbs, 3 RDF graph dbs, 10 Vector dbs, OpenSearch, Elasticsearch, Alfresco. 13 data sources (9 auto-sync), KG auto-building, RDF ontologies, schemas, LLMs, document processing, Docling, LlamaParse, GraphRAG, RAG only, Hybrid search, AI chat. React, Vue, Angular TypeScript frontends, FastAPI backend, REST API, MCP Server

Projet de détection de fake news utilisant une architecture Retrieval-Augmented Generation (RAG).

Le système compare un article avec une base de données vectorielle d’articles de référence afin de prédire un verdict (TRUE / FAKE / UNCERTAIN) et générer une justification avec un LLM local.

Pipeline :

Prétraitement et découpage du texte Génération d’embeddings (Ollama) Stockage dans ChromaDB Recherche des passages similaires Vote majoritaire pour le verdict Génération d’explication avec un LLM

Technologies : Python, Ollama, ChromaDB, scikit-learn, pytest.

Projet réalisé dans le cadre de la formation Développeur en Intelligence Artificielle (Simplon).

script generation, safety moderation, and content-quality filtering for 5-second vertical video creation

Full RAG pipeline (LangChain + Qdrant + Streamlit)

QuantForge AI is where intelligence is forged — not trained. A continuous engine that learns, adapts, and creates insight from the chaos of financial data.

--The alphabet is not language but a circuit of cognition.

Scholara: A Virtual Assistant Powered by LLM, RAG, and AI Agent.

Finding the right scholarship can be a timeconsuming and frustrating process. You'll likely spend hours poring over university websites, communicating with current students, and scouring the internet for information. Unfortunately, even after all that effort, there's no guarantee you'll find the perfect scholarship. That's where our AI scholar comes in. Let us handle the legwork for you, and we'll provide personalized assistance to find the most suitable scholarship for you and answer all your questions about the scholarships that interest you. We solved this by using LLM to provide a human-like when answer the students, RAG we build our own vector database to get more suitable scholarships from it based on student information and desire, also to keep our system up-to-date with all the updated information about the scholarships we built AI Web agent this helps the system to be aware about all the updates in the scholarships information, and new scholarships.

A dummy vector database

Docker build of postgresql changing the dimension limit for the cube extension, raising it to 350