bge

Multi-agent memory substrate for PostgreSQL — provenance-gated, vector-hybrid recall

Self-hosted knowledge base for AI agents: ingest Markdown into Qdrant with H1-H...-aware chunking, hybrid (dense+sparse) search and cross-encoder reranking.

Skill para o IA (claude, cursor, etc.) que permite consultar todas as APIs oficiais do IBGE (Instituto Brasileiro de Geografia e Estatística) — dados agregados (SIDRA), localidades, malhas geográficas, CNAE, nomes, notícias, indicadores de países, pesquisas e calendário.

AI academic Q&A chatbot for HUSC students, built with Flask, FastAPI, SQL Server, and Docker. It uses hybrid RAG with BGE-M3 embeddings, Qdrant/FAISS vector search, BM25 retrieval, RRF fusion, reranking, and Gemini to answer regulation questions with structured PDF parsing, citations, caching, intent detection, and rate limiting.

On-Premise RAG (продвинутая версия): vLLM/Ollama + bge-m3 + reranker + Qdrant + hybrid (BM25+RRF) + PDR + Basic Auth + Prometheus/Grafana + локальная оценка. Запускается на Ollama без GPU. Тех-задача Petrel AI (Astana Hub).

AI-powered Engineering P&ID Comparison and RAG Chatbot using FastAPI, ChromaDB, BGE-M3, and Qwen3.

MHAVE RAG Chatbot – Statistiques Haïti. Chatbot RAG (Retrieval-Augmented Generation) pour interroger des données statistiques haïtiennes. Utilise Ollama (nomic-embed-text, bge-m3), ChromaDB (base vectorielle), Pandas. Fichiers supportés : CSV/Excel.

Production-ready Retrieval-Augmented Generation (RAG) chatbot built with DeepSeek, LangChain, Hugging Face BGE-M3, and ChromaDB for intelligent document question answering.

Production-style RAG chatbot with hybrid retrieval (BM25 + BGE dense + cross-encoder reranking) and parent-child chunking over a 183-page policy PDF. Fully free stack — Groq, ChromaDB, sentence-transformers, Streamlit.

**RAG-Based AI Teaching Assistant:** Built an end-to-end RAG application that answers questions from educational videos using Whisper, BGE-M3 embeddings, semantic search, and Llama 3.2 via Ollama. Displays accurate answers with the relevant video title and timestamp through a Streamlit interface.

A fully local RAG-based Q&A system - hybrid BM25 + dense retrieval, BGE reranker, Ollama LLM, FastAPI backend, and a clean web UI. No API keys needed.

RAG-based YouTube chatbot that uses transcript retrieval, FAISS vector search, BGE-M3 embeddings, and Llama 3.2 via Ollama to answer questions about YouTube videos