CoreMLWin - Universal ML Runtime for Windows

🚀 Apple CoreML-like experience for Windows with DirectML GPU acceleration

✨ Features

🎯 Universal ML Inference - Run ONNX, PyTorch, TensorFlow models on Windows
⚡ GPU Acceleration - DirectML support for NVIDIA, AMD, and Intel GPUs
🔄 Automatic Fallback - CPU execution when GPU unavailable
🐍 Python SDK - Easy-to-use client library
🛡️ Production Ready - Security hardened, red-team tested
🔌 Simple Integration - Drop-in replacement for CoreML on Windows

📦 Installation

Option 1: Winget (Recommended)

winget install CoreMLWin.UniversalMLRuntime

Option 2: MSI Installer

Download from Releases and run:

msiexec /i CoreMLWin-0.1.0-x64.msi

Option 3: Portable (No Installation)

Download ZIP from Releases, extract, and run:

.\start-service.bat

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                    Your Application                         │
│                  (Python, C++, etc.)                        │
└─────────────────────┬───────────────────────────────────────┘
                      │
                      │ Client Library
                      ▼
┌─────────────────────────────────────────────────────────────┐
│              CoreMLWin Runtime Service                      │
│  ┌─────────────────────────────────────────────────────┐   │
│  │            Named Pipe IPC Server                    │   │
│  └──────────────────────┬──────────────────────────────┘   │
│                         │                                   │
│  ┌──────────────────────▼──────────────────────────────┐   │
│  │          Model Registry & Manager                   │   │
│  └──────────────────────┬──────────────────────────────┘   │
│                         │                                   │
│  ┌──────────────────────▼──────────────────────────────┐   │
│  │           ONNX Runtime Executor                     │   │
│  │  ┌─────────────┐  ┌─────────────┐  ┌────────────┐  │   │
│  │  │ DirectML    │  │     CPU     │  │   CUDA    │  │   │
│  │  │   (GPU)     │  │  Provider   │  │ Provider  │  │   │
│  │  └─────────────┘  └─────────────┘  └────────────┘  │   │
│  └─────────────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────────────┘
                         │
                         ▼
              ┌──────────────────────┐
              │   GPU / CPU          │
              │   Hardware           │
              └──────────────────────┘

🚀 Quick Start

1. Install Python SDK

pip install coremlwin

2. Start Service (if using portable)

coremlwin-service

3. Run Inference

from coremlwin_client import CoreMLWinClient
import numpy as np

# Connect to runtime
client = CoreMLWinClient()

# Check service health
print(client.health())

# Register a model
model_id = client.register_model("path/to/model.onnx")

# Prepare input
input_data = {
    "input": np.random.randn(1, 3, 224, 224).astype(np.float32)
}

# Run inference
output = client.run_inference(model_id, input_data)
print(output)

💡 Use Cases

🖼️ Computer Vision - Image classification, object detection, segmentation
📝 NLP - Text classification, sentiment analysis, translation
🎵 Audio Processing - Speech recognition, audio classification
🎮 Gaming - Real-time ML inference in games
🏢 Enterprise - Production ML deployments on Windows servers

📋 Requirements

Minimum

Windows 10 version 1903 (build 18362) or newer
Visual C++ Redistributable 2022
Python 3.8+ (for Python SDK)

🔧 Building from Source

Prerequisites

CMake 3.15+
Visual Studio 2019/2022
ONNX Runtime 1.16.3
Python 3.8+

Build Steps

# 1. Clone repository
git clone https://github.com/tropikandy/WinCoreML.git
cd WinCoreML

# 2. Download ONNX Runtime
$version = "1.16.3"
Invoke-WebRequest -Uri "https://github.com/microsoft/onnxruntime/releases/download/v$version/onnxruntime-win-x64-gpu-$version.zip" -OutFile "onnxruntime.zip"
Expand-Archive -Path "onnxruntime.zip" -DestinationPath "."
$env:ONNXRUNTIME_DIR = (Get-ChildItem -Directory -Filter "onnxruntime-win-*").FullName

# 3. Configure with CMake
cmake -B build -DCMAKE_BUILD_TYPE=Release -DONNXRUNTIME_DIR=$env:ONNXRUNTIME_DIR

# 4. Build
cmake --build build --config Release --parallel

# 5. Run tests
pytest tests/test_integration.py -v

📦 Distribution

For Developers

See RELEASE_CHECKLIST.md for release process.

One-command release:

.\tools\release.ps1 -Version 0.1.0

This will:

Run tests
Build MSI and portable packages
Create git tag
Trigger GitHub Actions
Auto-publish GitHub Release

For End Users

Multiple installation options available:

Method	Command	Best For
Winget	`winget install CoreMLWin.UniversalMLRuntime`	Most users
Chocolatey	`choco install coremlwin`	IT admins
MSI	Download from releases	Enterprise
Portable	Extract and run	No admin rights

See docs/INSTALLATION.md for details.

🛡️ Security

CoreMLWin has been thoroughly security tested:

✅ Red team audited - Comprehensive penetration testing
✅ All P0 critical fixes applied - Zero known critical vulnerabilities
✅ Input validation - Protection against buffer overflows, path traversal
✅ Secure hashing - SHA-256 content-based model identification
✅ 35+ security tests - Full test coverage for security features

See docs/RED_TEAM_AUDIT.md for the full security audit report.

📚 Documentation

Installation Guide - Detailed installation instructions
Distribution Guide - How to distribute your own builds
Release Checklist - Release automation guide
Security Audit - Security assessment report
Security Improvements - Implemented security fixes

🧪 Testing

Run All Tests

# Start service
start build\bin\Release\coremlwin_service.exe

# Run integration tests
pytest tests/test_integration.py -v

# Run security tests
pytest tests/test_security.py -v

# Run C++ security unit tests
.\tests\test_security_utils.exe

Test Coverage

✅ Integration Tests - End-to-end workflows
✅ Security Tests - Adversarial testing
✅ Performance Tests - Latency benchmarks
✅ Concurrency Tests - Thread safety
✅ C++ Unit Tests - Security utilities (35 tests)

📊 Performance

Latency Benchmarks (ResNet50, 224x224 input):

Provider	Latency	Throughput
DirectML (RTX 3080)	~5ms	200 FPS
CPU (Intel i9)	~45ms	22 FPS

Service Overhead: <100ms health check latency

🗺️ Roadmap

v0.2.0

Model caching with LRU eviction
Shared memory transport for zero-copy
Connection pooling
Google Test C++ unit testing framework

v0.3.0

WebSocket API support
REST API endpoint
Model versioning
A/B testing support

v1.0.0

Microsoft Store distribution
GUI management tool
Performance profiler
Model optimization tools

🤝 Contributing

Contributions are welcome! Please:

Fork the repository
Create a feature branch (git checkout -b feature/amazing-feature)
Commit your changes (git commit -m 'Add amazing feature')
Push to the branch (git push origin feature/amazing-feature)
Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

ONNX Runtime - High-performance ML inference engine
DirectML - GPU acceleration on Windows
Protocol Buffers - Efficient serialization
Microsoft - For the winget package manager

💻 Development & Disclaimer

CoreMLWin was developed using modern AI-assisted development practices. Architecture and design decisions by Andreas Larsson; implementation assisted by AI language models. All code has been reviewed, tested, and validated for production use.

Not affiliated with Apple Inc. CoreML is a trademark of Apple Inc.

📞 Support

🐛 Bug Reports: GitHub Issues
💬 Discussions: GitHub Discussions
📧 Email: Contact via GitHub profile

🌟 Star History

If you find CoreMLWin useful, please consider giving it a star ⭐

Made with ❤️ for the Windows ML community

GitHub • Install • Download • Report Bug

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
configs		configs
docs		docs
installer		installer
k8s		k8s
protos		protos
runtime		runtime
sdk		sdk
tests		tests
tools		tools
.gitignore		.gitignore
BUILD.md		BUILD.md
CONTRIBUTING.md		CONTRIBUTING.md
Dockerfile		Dockerfile
LAUNCH_NOW.md		LAUNCH_NOW.md
LAUNCH_READY.md		LAUNCH_READY.md
LICENSE		LICENSE
QUICK_START.md		QUICK_START.md
README.md		README.md
RELEASE_CHECKLIST.md		RELEASE_CHECKLIST.md
THIRD_PARTY_LICENSES.md		THIRD_PARTY_LICENSES.md
docker-compose.yml		docker-compose.yml

Latest commit

History

Folders and files

Repository files navigation