MLOS Linux (Ubuntu)

MLOS Linux Distribution - Ubuntu-based

A complete Linux distribution optimized for machine learning workloads, based on Ubuntu 22.04/24.04 LTS with MLOS Core, Axon, and kernel-level ML optimizations pre-installed.

Overview

MLOS Linux (Ubuntu) is a purpose-built Linux distribution that includes:

• ML-aware kernel scheduler (US-63/865,176) - Priority-based ML task scheduling
• Tensor memory management - Zero-copy tensor operations
• GPU resource orchestration - Multi-model GPU coordination
• Pre-installed MLOS Core - Kernel-level ML runtime (HTTP, gRPC, IPC APIs)
• Pre-installed Axon - Universal model installer (80%+ repository coverage)
• ML development toolchain - Python, PyTorch, TensorFlow, ONNX Runtime

Standards Compliance

• ✅ LSB (Linux Standard Base) - Interoperability compliance
• ✅ FHS (Filesystem Hierarchy Standard) - Standard directory structure
• ✅ Debian Policy Manual - Package format compliance
• ✅ systemd - Standard init system
• ✅ GPG Signing - Package and image signing

Architecture

MLOS Linux (Ubuntu) v1.0.0
├── Linux Kernel (6.x+ with MLOS patches)
│   ├── ML-aware scheduler
│   ├── Tensor memory management
│   └── GPU resource orchestration
├── Base System (Ubuntu 22.04/24.04 LTS)
│   ├── systemd (init system)
│   ├── NetworkManager (networking)
│   ├── Package manager (apt/dpkg)
│   └── Core utilities (LSB compliant)
├── MLOS Stack (.deb packages)
│   ├── mlos-core (v1.0.0+)
│   ├── axon (v1.5.0+)
│   └── mlos-toolchain
├── ML Development Toolchain
│   ├── Python 3.11+ with ML libraries
│   ├── CUDA toolkit (optional)
│   ├── PyTorch, TensorFlow, ONNX Runtime
│   └── ML development tools
└── Installation Tools
    ├── ISO installer (debian-installer/ubiquity)
    ├── Network installation (PXE)
    └── Cloud images (cloud-init)

IP Protection

This repository is public and contains only:

• ✅ Build scripts and configuration
• ✅ Distribution packaging files
• ✅ Installation system files
• ✅ Documentation

Private Components:

• 🔒 MLOS Core binaries (downloaded from private registry during build)
• 🔒 Kernel patches (downloaded from private registry during build)
• 🔒 Source code (remains in private repositories)

See IP Protection Plan for details.

Build Process

Prerequisites

• Ubuntu 22.04+ build environment
• GitHub Actions secrets configured:
  • MLOS_CORE_ARTIFACT_TOKEN - Access to MLOS Core artifacts
  • MLOS_KERNEL_PATCHES_TOKEN - Access to kernel patches
  • GPG_PUBLIC_KEY - Public key for signature verification

Build Steps

# 1. Clone repository
git clone https://github.com/mlOS-foundation/mlos-linux-ubuntu.git
cd mlos-linux-ubuntu

# 2. Download artifacts (requires tokens)
./scripts/download-artifacts.sh \
  --mlos-core-version v1.0.0 \
  --kernel-patches-version v1.0.0

# 3. Verify signatures
./scripts/verify-artifacts.sh

# 4. Build distribution
./scripts/build-ubuntu.sh \
  --mlos-core-deb mlos-core_1.0.0_amd64.deb \
  --kernel-patches kernel-patches/

# 5. Result: Bootable ISO image

Installation

ISO Installation

1. Download ISO from Releases
2. Create bootable USB: dd if=mlos-linux-ubuntu-v1.0.0.iso of=/dev/sdX
3. Boot from USB and follow installer
4. MLOS Core and Axon are pre-installed and auto-start on boot

Network Installation (PXE)

# PXE boot configuration
# See docs/INSTALLATION.md for details

Cloud Deployment

# Download cloud image
wget https://github.com/mlOS-foundation/mlos-linux-ubuntu/releases/download/v1.0.0/mlos-linux-ubuntu-v1.0.0-cloud.img

# Deploy to cloud provider
# See docs/CLOUD_DEPLOYMENT.md

Quick Start

After installation:

# MLOS Core is running
curl http://localhost:8080/health

# Install a model with Axon
axon install hf/bert-base-uncased@latest

# Register with MLOS Core
axon register hf/bert-base-uncased@latest

# Run inference
curl -X POST http://localhost:8080/models/hf/bert-base-uncased@latest/inference \
  -H "Content-Type: application/json" \
  -d '{"input": "Hello, MLOS!", "batch_size": 1}'

System Requirements

Minimum:

• CPU: 2 cores, x86_64 or ARM64
• RAM: 4GB
• Storage: 20GB
• GPU: Optional (NVIDIA, AMD, Intel)

Recommended:

• CPU: 4+ cores
• RAM: 8GB+
• Storage: 50GB+ SSD
• GPU: NVIDIA with CUDA support

Documentation

• Building Guide - How to build the distribution
• Installation Guide - Installation instructions
• Standards Compliance - LSB, FHS, Debian Policy
• IP Protection - IP protection strategy

Related Repositories

• mlos-linux-flatcar - Flatcar-based distribution
• mlos-linux-kernel - Shared kernel patches
• axon - Universal model installer
• core - MLOS Core runtime (private)

Status

Current Status: Planning Phase
Target Release: v1.0.0 - Q2-Q3 2026
Timeline: Phase 3 (Production Readiness)

Contributing

This repository is in planning phase. Contributions welcome once initial structure is established.

See CONTRIBUTING.md for guidelines (to be created).

License

• Distribution scripts/configs: MIT License
• MLOS Core binaries: Proprietary License (see LICENSES/MLOS_CORE_LICENSE.txt)
• Kernel patches: Proprietary License (see LICENSES/KERNEL_PATCHES_LICENSE.txt)
• Base system (Ubuntu): Ubuntu license (as-is)

---

MLOS Foundation - Building the future of ML infrastructure.