AI-103 · Azure AI Apps and Agents Developer Associate
AI-103: Implement computer vision solutions (10–15%)
Computer vision represents 10–15% of AI-103 and now spans generation, editing, multimodal understanding, accessibility and safety.
Generation and understanding
Know when to use text-to-image or video generation, reference media, masks and inpainting. Multimodal models can analyze visual context, answer questions grounded in images, produce captions and create accessible alt text.
Visual safety
Moderate unsafe media before processing, detect indirect prompt injection embedded in images and enforce policy such as watermark or prohibited-symbol requirements. Prompt controls alone do not replace image-specific screening.
Official Microsoft source