High-performance late-interaction retrieval engine for on-prem AI. ColBERT/ColPali multi-vector search with Rust fused MaxSim, Triton GPU kernels, ROQ quantization, LEMUR routing, WAL-backed CRUD, and a FastAPI server — single machine, CPU or GPU.

High-performance database management system

This repository is dedicated to my learning journey through the LangChain & Vector Databases in Production course offered by Activeloop. The course is part of the Gen AI 360 Foundational Model Certification and focuses on mastering Large Language Models (LLMs) and Vector Databases.

Production-ready enterprise multi-agent AI orchestration platform with distributed memory architecture, intelligent task routing, semantic search, and comprehensive monitoring using modern software architecture patterns.

Air-gapped, production-ready RAG workspace featuring hybrid search, semantic reranking, cross-encoder reranking, and verifiable citations.

An offline RAG system that bridges Vietnamese-English medical knowledge. Ask questions in Vietnamese, get answers from English medical literature—translated back to Vietnamese. Built for doctors, students, and health enthusiasts who need quick access to medical information while maintaining complete data privacy.

RAGenix is a production-ready RAG system that enables accurate, context-aware conversations with PDFs using hybrid search, reranking, and conversational memory.

A comprehensive, hands-on tutorial repository for learning and mastering LangChain - the powerful framework for building applications with Large Language Models (LLMs). This codebase provides a structured learning path with practical examples covering everything from basic chat models to advanced AI agents, organized in a progressive curriculum.

Agentic RAG with Reasoning is a sophisticated AI system built with Agno, Gemini, and OpenAI that provides transparent, step-by-step insights into its research process. It enables users to query web sources while observing the agent's real-time thought process and logical reasoning steps.

RAGKNO — A production-grade RAG system with semantic reranking, hybrid retrieval, multi-source ingestion (Google Drive, PDF, URL), persistent multi-session memory, and source-cited answers. Built with FastAPI, FAISS, and Google Generative AI.

This project is a Streamlit-based AI search engine that uses Agentic Retrieval-Augmented Generation (RAG) powered by LangGraph and Google Gemini. It allows users to input a blog URL, which it crawls and indexes in a vector database, to intelligently answer specific questions based on the article's content.

Comprehensive medical imaging management system with a DICOM viewer, HU conversion, 28 image filters, and dual database architecture (Qdrant + SQLite). Supports radiology report generation, patient data management, and vector similarity search.

A practical and critical evaluation of Retrieval-Augmented Generation (RAG) systems on legal/insurance documents using RAGAS. This project analyzes metric failures, false negatives, retrieval pitfalls, and proposes a more realistic composite evaluation score.

communication between agent using kafka messaging queue

Hệ thống RAG báo cáo tài chính Việt Nam hỗ trợ OCR, hybrid search, semantic indexing, chống trùng lặp SHA-256 và suy luận bằng local LLM qua Ollama.

Transform your legal document review from days to minutes with LegalRAG. Our AI assistant understands complex legal questions, searches thousands of documents with semantic precision, and delivers professional answers with source citations. Experience faster case prep and thorough due diligence.

AI Chatbot using Vector Database (FAISS) with semantic search