⚡ MedMemory AI – Core Backend Server Engine

An open-source, AI-native clinical intelligence engine. Track, parse, and map longitudinal patient medical history locally.

This repository houses the high-performance local RAG backend server engineered with FastAPI, LangChain, LangGraph, Qdrant Vector DB, and PostgreSQL, completely powered by offline Ollama models.

---

📚 Central Documentation Portal

For enterprise architecture blueprints, detailed multi-agent workflows, database designs, and front-end interface integrations, please visit our centralized tracking zones:

• Install App APK: Latest Release
• 🌐 Live Documentation Portal: medmemory-ai.github.io/docs
• 🏛️ Global Organization Workspace: @MedMemory-AI

---

🛠️ Tech Stack Primitives (Backend Specific)

• API Framework: FastAPI (Asynchronous request handling)
• Orchestration Framework: LangChain & LangGraph (Stateful multi-agent cycles)
• OCR & Layout Parser: Docling (Structured JSON extraction)
• Local LLM Engine: Ollama (llama3.2 & mxbai-embed-large)
• Relational Database: PostgreSQL (Chronological metrics & schemas)
• Vector Database: Qdrant DB (Dense spatial vector search metrics)

---

🚀 3-Step Local Quick Start

If you are a contributor working specifically on the Python codebase, you can fire up the local development engine using these steps:

Detailed Guide: Local Setup Guide

---

1. Initialize Your Environment

git clone [https://github.com/MedMemory-AI/server.git](https://github.com/MedMemory-AI/server.git)
cd server
python -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`
pip install -r requirements.txt

2. Boot Local Infrastructure Layers

Ensure your local Ollama instance is active and pull the model signatures:

ollama pull llama3.2
ollama pull mxbai-embed-large

Spin up your local storage instances via Docker:

docker run -d -p 6333:6333 qdrant/qdrant
docker run -d -p 5432:5432 -e POSTGRES_PASSWORD=local_secret postgres

3. Run the Backend Server

uvicorn app.main:app --reload --port 8000

The API interactive sandbox environment will immediately be available at http://localhost:8000/docs.

---

🤝 How to Contribute

Before submitting a Pull Request, please read our centralized Contributing Guidelines. Ensure you match local code-style rules and always execute your task variations on a dedicated feature branch (feat/your-patch).