Two clouds
Same service twice
One reference app on two providers, differences named.
Multi-Cloud AI Deployment
AWS, Azure, GCP — managed model APIs, identity, networking, cost, and an exit plan so one cloud is a choice, not a trap.
Created by Baljeet Dogra
Course Objectives
- Deploy the same shape of service on two clouds.
- Explain identity, private networking and keys without a vendor diagram.
- Cost the same workload in two places with honest numbers.
- Write the reverse conditions for the cloud you picked.
What you will produce
ADR
Cloud decision
Rejected options, residency, exit, cost.
4 weeks curriculum
Expand a part for the syllabus. Content stays searchable when closed.
01 The map Week 1
Managed model APIs, regions, quotas, what is portable.
- Bedrock
- Azure OpenAI
- Vertex
- Quotas
02 Identity and path Week 2
IAM, private endpoints, keys, audit logs.
- IAM
- Private link
- Keys
- Audit
03 Cost and residency Week 3
Bills, reservations, data residency, training rights.
- Cost
- Residency
- Rights
- FinOps
04 Exit Week 4
Second cloud, abstraction cost, capstone ADR.
- Portability
- Abstractions
- ADR
- Capstone
Who this is for
Platform and enterprise engineers
Procurement picked two clouds. You have to make that true.
Architects
You write the paper. You should have deployed twice.
Prerequisites
- You have deployed on at least one major cloud
- Containers at a basic level
- IAM literacy
Not a fit if
- A certifications tour of three consoles
- People with no cloud account experience
Related: Kubernetes for AI Systems · Enterprise AI Engineer
Questions
Do I need three paid accounts?
Two is enough for the capstone. The third is mapped, not necessarily deployed.
Open weights or managed APIs?
Both appear. Managed APIs are the enterprise default; portability includes an open path.
Ready to start?
Four weeks to the same service on two clouds, with an exit. Create an account to enrol.
Enrol now