AI-103 · Azure AI Apps and Agents Developer Associate
AI-103: Plan and manage an Azure AI solution (25–30%)
This area is 25–30% of AI-103. Translate requirements into services, deployment boundaries, identity, observability and governance before writing application code.
Architecture decisions
Compare large, small and multimodal models; grounding, vector search and agent services; Global, Data Zone and regional deployment boundaries; Standard and provisioned capacity. Requirements for residency, throughput, latency and predictable cost often point in different directions.
Operate securely
Use managed identity, keyless credentials, private networking and least-privilege roles. Monitor quotas, rate limits, cost, drift, safety events, ingestion quality and search relevance. Treat trace logging, provenance and approval workflows as governance controls, not optional debugging extras.