Chiron

Build AI applications on any major cloud.

Chiron is a public, no-auth developer guide for building production AI services on Azure, Google Cloud, or AWS — running on Kubernetes or on each provider's serverless runtime.

Every recipe on this site follows the same shape:

1. Pick your provider (Azure / GCP / AWS) once — the choice syncs across every tabbed block on the page.
2. Pick your compute model (Kubernetes / Serverless) per section.
3. Copy the Terraform → Helm → Python → verify blocks.

Foundations landed in E1 (auth, compute choice, the reusable Terraform + Helm baseline, local dev setup). E2+ layers real recipes on top.

Why Chiron

Cloud vendor docs each show you their pattern. Chiron shows the same pattern side-by-side across all three, so choosing a provider becomes a scoped decision — not a re-learn.

What Chiron does not cover

• Vendor-specific proprietary features that don't have a peer on the other clouds. If it's Azure-only or AWS-only, it doesn't belong here — cross-provider portability is the spine.
• Anyone's private infrastructure. Every example targets a public cloud service, so every reader can run it end to end.

License

MIT — build whatever you want. If a recipe helps, a link back to chiron.tonyvigna.com is appreciated but not required.