Helios

A phased operating systems project that evolves a Linux-hosted object storage daemon into a kernel-backed platform.

Published on March 01, 2026 by Matthew Hambrecht

helios operating-systems rust architecture

1 min READ

Helios is my long-horizon systems project: build an operating system and its core services from scratch, but do it incrementally in a way that keeps real functionality usable at each phase.

Instead of jumping straight into kernel-only development, Helios will start with practical components running on Linux and then migrates those same components toward a custom kernel target.

Architecture at a glance

Helios Phase 0 Architecture

The Phase 0 architecture centers on four components:

  • helid: object storage daemon and control plane core
  • helictl: CLI client for remote control and operations
  • helios-http: HTTP interface layer for daemon interaction
  • helios-sci: the system call interface abstraction that decouples higher-level services from the concrete OS target

Why this design

The key idea is to isolate platform-specific behavior behind Helios SCI so daemon and service logic are not tightly coupled to Linux syscalls or one runtime model.

That lets me:

  • iterate faster on storage and control-plane behavior early
  • validate APIs and workflows before deep kernel work
  • migrate components progressively instead of rewriting everything at once

Current phase and roadmap direction

Phase 0 is about core architecture and a mock syscall interface with modular daemon/CLI/API boundaries. The next milestone track will move through:

  1. booting a simple kernel
  2. freestanding/no-std execution paths
  3. bringing daemon capabilities onto the kernel-backed environment

The result I’m targeting is a system where the same high-level service model can operate across both Linux-hosted and custom-kernel-backed execution.

Follow the project