← All certifications
AI-103 · ASSOCIATE

Azure AI Engineer — Developing AI Apps and Agents

The associate-level builder exam, replacing AI-102. Centered on Azure AI Foundry as the unified development platform, with heavy emphasis on generative AI and agentic solutions, RAG pipelines, Azure AI Search, computer vision, and information extraction. Assumes hands-on Python development.

$165
Exam fee
100
Minutes
40–60
Questions
700
Pass / 1000
1 yr
Renews
⚡In beta. In beta as of 2026 — replaces AI-102 (which retires June 30, 2026). Beta exams may have a reduced fee and delayed scoring. Objectives reflect the April 2026 study guide and emphasize Foundry, agents, RAG, and Content Understanding far more than AI-102 did.
Exam objectives — check off as you master each domain
Plan and manage an Azure AI solution
25-30%
No PowerStack tutorials mapped yet — coming soon.
Implement generative AI and agentic solutions
30-35%
No PowerStack tutorials mapped yet — coming soon.
Implement computer vision solutions
10-15%
No PowerStack tutorials mapped yet — coming soon.
Implement text analysis solutions
10-15%
No PowerStack tutorials mapped yet — coming soon.
Implement information extraction solutions
10-15%
No PowerStack tutorials mapped yet — coming soon.
Free preview — no account needed

Sample AI-103 practice questions

5 of the 60 original questions in our AI-103 bank, written to match the real exam's style and difficulty. Every question includes a full explanation — expand each answer to check your reasoning.

  1. Question 1medium

    When choosing a model for a task, which factors should an AI engineer weigh?

    • AModel type (LLM, small language model, multimodal), capability, cost, and latency
    • BOnly the model's release date
    • COnly the vendor logo
    • DOnly the model name length
    Show answer & explanation
    Correct: A

    Weigh model type (LLM/SLM/multimodal), capability, cost, and latency.

  2. Question 2medium

    Which is the unified platform for building, deploying, and managing AI apps and agents in this exam?

    • AMicrosoft Foundry
    • BMicrosoft Excel
    • CAzure DNS
    • DWindows Update
    Show answer & explanation
    Correct: A

    Microsoft Foundry is the unified platform for building/deploying/managing AI apps and agents.

  3. Question 3medium

    For an agent that must look up enterprise data, which retrieval approach is appropriate?

    • AChoose an appropriate retrieval/indexing method (e.g., vector or hybrid search)
    • BHard-code all answers
    • CDisable retrieval
    • DUse a print spooler
    Show answer & explanation
    Correct: A

    Choose an appropriate retrieval/indexing method such as vector or hybrid search.

  4. Question 4medium

    Which security configuration is recommended for AI app resources?

    • AManaged identity, keyless credentials, private networking, and role policies
    • BShared passwords in source code
    • CPublic endpoints with no auth
    • DDisabling RBAC
    Show answer & explanation
    Correct: A

    Use managed identity, keyless credentials, private networking, and role policies.

  5. Question 5medium

    Which is a sound choice for grounding generative tasks on trusted data?

    • AUse Foundry services for grounding and vector search
    • BIncrease temperature to maximum
    • CRemove all data sources
    • DUse a DNS lookup
    Show answer & explanation
    Correct: A

    Use Foundry services for grounding and vector search to base answers on trusted data.

Ready for the full mock exams?

60 questions with per-domain score breakdowns, a timer, and saved progress. Free account — no card, no paywall.

AI-103 Exam Prep, Practice Questions & Study Guide — PowerStack