Skip to content

blindml-ai/blindml-server

Repository files navigation

BlindML Server

BlindML is an open-source MLOps platform powered by Fully Homomorphic Encryption (FHE)
It enables secure, lightweight, and fast deployment of AI models — without ever decrypting your data.

✨ What is BlindML Server?

BlindML Server is a Node.js backend for running AI inference on encrypted data using FHE (Fully Homomorphic Encryption).
It supports secure model inference through a simple API, without decrypting user data.
Uploaded models are automatically deployed in Docker containers for easy management and scaling.

📡 Model inference is requested via BlindML Client

🚀 Key Features

  • Privacy-preserving AI inference using Fully Homomorphic Encryption
  • Lightweight and fast server optimized for encrypted data handling
  • Easy integration with Python clients via BlindML Client
  • Developer-friendly: simple to deploy, easy to customize
  • Works seamlessly in secure environments where data must remain encrypted

🖥️ Supported Environments

OS Support
Windows ❌ Not supported
macOS ❌ Not supported
Linux ✅ Fully supported

Note:
If Docker is installed on Linux and communicate with blindml-server remotely, it can also be used on Windows and macOS environments.

🔐 Supported FHE Libraries

BlindML Server supports multiple FHE libraries for encrypted inference:

Library Description
concrete-ml (by Zama) Open-source FHE library
CSEM (soon) A fast FHE inference engine that supports 100+ machine learning functions, optimized for performance and flexibility.

After starting the BlindML server, you can choose the FHE library to use when deploying your model.


🧰 Getting Started

Prerequisites

Make sure you have the following installed:

Edit your .env file

To run this server, edit the .env file in the root directory. If the .env file is not modified, the application will run with the following default configuration."

# Service server configuration
SERVICE_SERVER=http://127.0.0.1
SERVICE_PORT=3000

# Docker server configuration
DOCKER_SERVER=http://127.0.0.1
DOCKER_PORT=80

# Authentication server configuration
AUTH_SERVER=http://127.0.0.1
AUTH_PORT=80

# Database configuration
# Leave these empty to use SQLite3 (default).
# If values are provided, the project will use MySQL.
DB_HOST=
DB_USER=
DB_PASSWORD=
DB_NAME=

⚙Installation

# 1. Clone the repository
git clone https://github.com/blindml-ai/blindml-server.git
cd blindml-server

# 2. Install dependencies
npm install

# 3. Start the server
node server.js

About

BlindML Server is a backend server that enables the deployment and inference of AI models using Fully Homomorphic Encryption (FHE)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors