awslabs

awslabs / global-capacity-orchestrator-on-aws

Public

GCO is an experimental platform that spins up EKS Auto Mode clusters across AWS regions, wired together with Global Accelerator for low-latency routing. It handles the heavy lifting of multi-region compute orchestration — capacity-aware scheduling, spot fallback, globally distributed autoscaling inference — and offers a REST API, CLI and MCP server

14
2
100% credibility
Found May 01, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Global Capacity Orchestrator on AWS enables multi-region orchestration of GPU-accelerated workloads like AI training and inference through a unified CLI and API.

How It Works

1
🔍 Discover easy global compute

You hear about a tool that lets you run AI and GPU jobs across the world without managing servers.

2
📦 Get the simple tool

Install the friendly command-line helper in seconds so you can start using it right away.

3
🚀 Launch your worldwide setup

With one command, set up smart clusters everywhere that find capacity and handle jobs automatically.

4
📊 Find the best spot and run a job

Check where resources are ready, send your first task, and see it start working.

5
📱 Watch progress and grab results

Follow along with updates, view logs, and download your finished work safely stored.

6
🤖 Add smart AI services

Launch prediction services that spread across places, scale as needed, and switch if one slows down.

🎉 Global jobs running smoothly

Your AI tasks now run reliably anywhere with savings, backups, and no hassle.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 14 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is global-capacity-orchestrator-on-aws?

This experimental Python platform deploys EKS Auto Mode clusters across AWS regions, connecting them via Global Accelerator for low-latency routing and automatic failover. It automates multi-region compute orchestration with capacity-aware scheduling, spot fallback, and distributed autoscaling inference endpoints—all accessible through a single REST API and CLI. Users submit Kubernetes manifests or deploy inference servers once, and it handles capacity discovery, provisioning, and global load balancing without per-cluster management.

Why is it gaining traction?

One-command deploy and teardown spins up full multi-region setups with IAM auth, skipping kubeconfig hassles and manual scaling. Capacity-aware routing picks available accelerators across regions, with spot fallback for cost savings and autoscaling for inference workloads. Built-in observability, cost tracking, and demos make testing fast, standing out from fragmented EKS tools.

Who should use this?

ML engineers training LLMs or running batch inference needing redundancy without ops toil. HPC teams chasing GPU capacity across regions. Devs prototyping global compute who want CLI-driven autoscaling over manual cluster wrangling.

Verdict

Worth a spin for experimental multi-region EKS setups—demos and docs are polished, CLI is intuitive. But with 14 stars and 1.0% credibility score, treat as alpha: test in sandboxes before production.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.