Retrieval and grounding with Azure AI Search
AI-103 · Lesson 14 of 15 · 1:46:41
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the AI-103 exam (Azure AI Apps and Agents Developer Associate). In this lesson: ingesting and indexing content with Azure AI Search (data sources, indexers, the push API, normalized images, audio and video through Content Understanding), enriching with skills (OCR and Text Merge, Document Layout, the Content Understanding skill, GenAI Prompt, Custom Web API, integrated vectorization and index projections), vector search (HNSW vs exhaustive KNN), hybrid search with Reciprocal Rank Fusion and the semantic ranker, a hybrid + semantic query in Python, and connecting retrieval to agents (the Azure AI Search tool, file search, Foundry IQ knowledge bases, Search roles). Includes three original practice questions.
Preview the topics
- A pipeline indexes PDF reports. Searches find the typed text, but never the text inside the charts and photos embedded in the pages. What’s missing?
- Support engineers search manuals by error code, like E-4012. Pure vector search returns manuals about similar errors, but not the one with that exact code. What should you use?
- An HR agent must answer from an existing Azure AI Search index that another team maintains, and cite its sources. What should you add to the agent?