The Azure generative AI stack, end to end, for engineers and architects: Azure AI Foundry as the front door, Azure OpenAI and the model catalog for models, Provisioned Throughput and Batch for the bill, plus the retrieval, agent, safety, cost and governance layers that turn a model into a product. A 30-part series that reads from first principles to production. Where it meets vendor-neutral ground it links to the Generative AI guide and the AWS Gen AI guide rather than repeating them.
Series in progress · 20 of 30 published
Phase 1 · Platform foundations
- 01What the Azure GenAI Stack Is, End to End
- 02The Azure AI Foundry Platform
- 03Azure OpenAI in Foundry Models
- 04The Model Catalog and Models-as-a-Service
- 05Azure OpenAI vs Foundry vs Azure ML
- 06Deployment Types: Standard, PTU and Batch
- 07ND-Series GPUs and Maia
- 08Regions, Quotas and Data Zones
- 09Private Link and VNet for Azure OpenAI
- 10Entra ID, Managed Identities and Encryption
Phase 2 · Calling models, RAG and agents
- 11Calling Models: REST, SDKs and the Responses API
- 12Azure AI Search and RAG
- 13Azure AI Foundry Agent Service
- 14Azure AI Content Safety
- 15Prompt Flow
- 16Fine-Tuning Azure OpenAI
- 17Distillation and Stored Completions
Phase 3 · Open models, training and multimodal
- 18Open Models on Azure ML
- 19Distributed Training on ND Clusters
- 20Data Prep, Grounding and Indexing
- 21Semantic Kernel, AutoGen and Agent Framework
- 22Microsoft Copilot Studio
- 23Evaluation and Observability in Foundry
- 24Multimodal: Vision, Audio, Image and Docs
Phase 4 · Operations, cost and governance
- 25Azure Monitor Observability
- 26Cost Governance and FinOps
- 27Responsible AI and the RAI Dashboard
- 28LLMOps and CI/CD on Azure
- 29Reference Architectures on Azure
- 30Azure GenAI vs the Field, the Verdict
DrJha