AI Agent · RAG · Multimodal Document Understanding

Hi, I'm Lu Minghao.

I'm a Computer Science student at Shanghai University, focusing on AI Agents, RAG systems, multimodal document understanding, and LLM application engineering.

About Me

I enjoy building practical AI systems that combine retrieval, memory, model evaluation, and backend engineering. My current interests include long-term memory for LLM applications, hybrid retrieval, document parsing, OCR, and multimodal model applications.

Featured Projects

Feishu Project Decision Memory System

A long-term memory system for project discussions. It extracts decisions, reasons, owners, deadlines, and historical context from Feishu messages and documents.

CLI Long-Term Memory System

A command-line memory tool for developers, supporting command recording, workflow reuse, natural language search, and context-aware suggestions.

Long-Memory Enhanced RAG System

A multi-stage RAG system with hybrid retrieval, reranking, active retrieval, multi-hop retrieval, and memory gating.

Document Understanding System

An end-to-end document parsing pipeline that converts images and PDFs into structured Markdown and JSON using layout analysis, OCR, and formula detection.

Skills

Python C/C++ FastAPI PyTorch Transformers MindSpore RAG FAISS ChromaDB OCR Multimodal Models SQLite SQLAlchemy Docker Linux

Latest Blog Posts

How I Built a Long-Term Memory System for AI Agents

Coming soon.

Hybrid Retrieval in RAG: Vector Search + Keyword Search

Coming soon.

Building an End-to-End Document Understanding Pipeline

Coming soon.

Contact

Email: lmh04102@163.com

GitHub: github.com/LuMH1027

Phone: +86 15901639546