Skip to main content

Overall Platform Architecture designed

This is the top-down architecture ladder for AI Cloud. A reviewer should be able to start here, understand the product shape, then choose the right next packet without reverse-engineering the repo.

The System In One Sentence​

AI Cloud is a shared control plane for GPU runtimes, apps, and future platform products, with reusable IAM, billing, policy, evidence, registry, and runtime access services underneath product-specific workflows.

The Architecture Ladder​

What Each Layer Answers​

LayerMain questionPrimary page
VisionWhat are we building and why is it a platform, not just one app?Architecture and Design Principles
Shared modelWhich capabilities are shared across products vs product-owned?Platform Shared Services
Runtime topologyHow does traffic, control, async work, and node access actually flow?System Overview
Code/layer ownershipWhere does code belong and how is deployment separation supposed to evolve?Code Structure And Layer Model
Deep engineeringWhat is the detailed model for each major service family?Detailed Design Index
Product realityHow does this show up in user, admin, operator, and release workflows?Use AI Cloud, Operators, Security & Production Readiness

Platform Shape​

Core Design Claim​

The repo is not organized around “frontend vs backend” or “one service per feature.” It is organized around a stable control-plane model:

  • contracts first;
  • shared business authority in platform services;
  • product-specific behavior in product domains;
  • provider and environment specifics behind adapters and config;
  • production as a profile of the same system, not a different product.

What Is Easy To Miss​

The codebase already contains more than a thin provisioning app:

  • shared IAM and access posture;
  • immutable billing and metering model;
  • audit, evidence, and release-readiness posture;
  • terminal and workload runtime access surfaces;
  • app-platform and SDK hooks;
  • environment and deployment thinking that reaches beyond local demo mode.

That is why the architecture section needs both overview pages and deep service packets.

  1. Architecture and Design Principles
  2. Overall Platform Architecture
  3. System Overview
  4. Code Structure And Layer Model
  5. Platform Shared Services
  6. Detailed Design Index

Use This Page As The Jump Point For Reviews​

Canonical sources