Noureddine RAMDI / Olares: Building a personal cloud OS on Kubernetes for local AI and app hosting

Created Wed, 06 May 2026 18:58:37 +0000 Modified Sat, 23 May 2026 20:41:27 +0000
1169 Words
github-stars go kubernetes personal-cloud self-hosted ai infrastructure

beclab/Olares

Olares takes the concept of public cloud infrastructure and maps it onto local hardware, creating a full personal cloud operating system. It runs Kubernetes (via K3s) under the hood, integrating a vast array of open-source projects to deliver a cloud-like experience on your own device. This means users can run isolated apps, manage storage that scales automatically, and handle AI workloads locally — all without relying on third-party clouds.

A personal cloud OS built on Kubernetes and CNCF projects

At its core, Olares is a self-hosted personal cloud operating system written in Go. It leverages K3s, the lightweight Kubernetes distribution, as the foundational orchestration layer. This choice is key: Kubernetes provides the infrastructure-as-a-service (IaaS) primitives, container scheduling, and network management needed to reliably run multiple isolated workloads on commodity hardware.

But Olares goes beyond just Kubernetes. It integrates over 30 third-party open-source components, many from the Cloud Native Computing Foundation (CNCF) ecosystem, stitching them together into a unified platform. Some notable components include JuiceFS for scalable distributed filesystem support, MinIO for object storage, Envoy as a service mesh and proxy, Authelia for unified single sign-on (SSO) with enterprise-grade security, and Tailscale for secure networking.

This results in a system that provides local equivalents of public cloud services — infrastructure, platform, and software layers — but all running on your own machine or network. The platform ships with built-in apps like a file manager, vault for secrets, a reader, a marketplace, and a customizable dashboard. Dedicated clients are available for desktop, mobile, and browser access.

One of the standout features is built-in support for local AI workloads. Olares integrates Ollama and ComfyUI, enabling users to run large language models and AI interfaces locally, offering data sovereignty even for AI applications. This is particularly relevant as AI workloads grow in demand but cloud privacy concerns rise.

Technical architecture and the tradeoffs it involves

Olares’ architecture is ambitious. Using Kubernetes (K3s) as the base provides a stable, battle-tested orchestration layer, but it also means the platform inherits Kubernetes’ complexity and resource requirements. While K3s is lightweight compared to full Kubernetes, it still demands a certain level of hardware capability and operational knowledge.

The integration of 30+ open-source projects is both a strength and a challenge. On one hand, it leverages mature, community-vetted software for storage, networking, security, and application hosting. On the other, maintaining coherence and smooth interoperability among so many moving parts requires careful design and ongoing maintenance.

Code quality is generally high, with Go used for core components ensuring performance and concurrency control. The platform’s approach to sandboxed application isolation enhances security by limiting app privileges and attack surfaces. Unified SSO via Authelia simplifies user management and access control across hosted services.

Storage is handled with automation and scalability in mind. JuiceFS enables distributed, scalable filesystem capabilities, while MinIO provides object storage compatible with AWS S3 APIs. This means users can expect cloud-like storage behavior locally, including automated scaling and redundancy.

Networking is robust, with Envoy managing service mesh functions like load balancing and traffic routing, and Tailscale providing zero-config VPN capabilities to securely connect devices and services. These choices reflect an enterprise-grade approach to self-hosted cloud infrastructure.

The inclusion of local AI workload support is noteworthy. Integrating Ollama and ComfyUI directly into the platform allows AI inference and interaction without sending data to external clouds. This opens paths for privacy-conscious users and developers wanting to experiment with AI models locally.

The tradeoff is clear: Olares offers a comprehensive personal cloud OS but comes with the complexity of Kubernetes and a multi-component stack. It’s not a zero-config plug-and-play system for casual users. Hardware requirements and operational overhead may be significant compared to simpler NAS solutions.

Explore the project

Olares targets Linux distributions, specifically Ubuntu 24.04 LTS or later and Debian 11 or later. The project’s source code and documentation are on GitHub at https://github.com/beclab/Olares.

The repository is primarily in Go, reflecting its focus on performance and concurrency. Documentation includes a Getting Started Guide, which outlines step-by-step instructions for setting up the system on compatible Linux hosts.

While there isn’t a quick CLI install snippet in the README, the docs guide users through installation and configuration, emphasizing prerequisites and system compatibility. The modular design encourages exploration through the codebase and documentation to understand how the various integrated projects are configured and orchestrated.

Users interested in extending Olares or integrating their own apps will find the modular architecture and Kubernetes foundation a flexible base to build on. The built-in apps demonstrate how to leverage the platform’s APIs and services.

Verdict: Who should consider Olares?

Olares is a compelling personal cloud OS for users who want cloud-like capabilities on their hardware and are comfortable with Kubernetes and Linux systems. It fits well for homelab enthusiasts, privacy-conscious AI developers, and those seeking a NAS alternative focused on application hosting, local AI, and decentralized social media.

It’s not the easiest system to set up or manage — the layered architecture and dependencies on many CNCF projects mean operational overhead and potential complexity. But for its target audience, it offers a uniquely integrated platform that balances security, scalability, and functionality.

If you want a self-hosted cloud platform that supports sandboxed apps, unified SSO, automated scalable storage, and local AI workloads — all built on a solid Kubernetes foundation — Olares is worth evaluating. Just be prepared for the learning curve and hardware needs that come with running a full personal cloud OS stack.

# Getting started

# System compatibility
# Olares has been tested and verified on the following Linux platforms:
# - Ubuntu 24.04 LTS or later
# - Debian 11 or later

# Set up Olares
# To get started with Olares on your own device, follow the Getting Started Guide for step-by-step instructions.

Overall, Olares is a technically interesting project that pushes the boundary of what personal cloud OSes can achieve by directly adopting and integrating cloud-native components on local hardware. Its honest tradeoffs and ambitious scope make it a worthwhile study for advanced users and developers in the self-hosting space.

→ GitHub Repo: beclab/Olares ⭐ 4,512 · Go