Microsoft AI-102 Exam Dumps

Question 5:
You’re right to distinguish assignability from who receives the license.
The correct answer should be Group1, Group2, and Group3 only—not all five.

• Group1: Security group; eligible.
• Group2: Security group; eligible.
• Group3: Security group; eligible, even though it currently contains only Device1. The license assignment can be made to the group, but no device receives a user license.
• Group4 and Group5: Microsoft 365 groups; they are not supported for group-based license assignment in this scenario.

then what about the first objective? it says finalize .... and migration plan
You’re right to question it. If the first objective explicitly says “finalize the Implementation and Migration Plan,” it belongs to Phase F, not Phase E.
The correct distinction is:

• Phase E – Opportunities & Solutions: Identifies major implementation projects, Transition Architectures, and the overall migration strategy.
• Phase F – Migration Planning: Finalizes the Implementation and Migration Plan, including prioritization, costs, risks, and sequencing.

• 1 ? F: Finalize the migration plan
• 2 ? F: Confirm readiness and ability to undergo change
• 3 ? E: Determine the need for an incremental approach and identify Transition Architectures
• 4 ? G: Perform governance during implementation

Question 52:
Question 52’s stated answer looks partly incorrect.

• Commit trigger: This is appropriate for automatically starting an image build when source code is committed. However, because the scenario also says images must rebuild when a Docker Hub base image changes, a base-image-update trigger is also relevant.
• Build service: The requirement explicitly says image build automation must run inside Azure Container Registry and eliminate external build services. Therefore, the correct service should be ACR Tasks, not a GitHub workflow.

• Build container images in Azure, without a developer machine.
• Trigger builds from source-code commits.
• Rebuild images when a base image is updated.
• Push the resulting image to ACR.

• Trigger: Commit (and possibly Base image update for the rebuild requirement)
• Build mechanism: ACR Task

Question 17:
Question 17 tests how an Azure Container App securely accesses Azure Key Vault without storing credentials.
The answer key’s sequence appears questionable. The logical sequence should be:

1. Assign a system-assigned managed identity to the Azure Container App

1. Grant the managed identity the Key Vault Secrets User role

1. Retrieve the secret by using the Azure SDK

Question 12:
Question 12 is a hotspot question, but the statements and answer choices are shown only as [Image], so I can’t see what it is testing or verify the answer key.
The reference to Microsoft’s privacy page suggests it may involve concepts such as:

• Privacy by design
• Transparency about data collection and use
• User control over personal data
• Data protection and security
• Compliance with applicable privacy laws

Configure AWS Security Hub This is optional and only needed for importing Security Hub findings. It is not required for onboarding or agent deployment.
You’re right. The answer key’s B is incorrect for this wording.
Correct answer: D — Create an Access control (IAM) role for Defender for Cloud.
For onboarding an AWS account through the native Defender for Cloud connector, Defender for Cloud needs an AWS IAM role with the required trust relationship and permissions. This role allows Azure to access the AWS account securely without using long-term AWS user credentials.

• AWS Security Hub: Optional. It is required only if you want to import Security Hub findings into Defender for Cloud.
• AWS SSM agent: Used for certain Defender for Servers capabilities, but it is not the first onboarding prerequisite.
• AWS user: Not the recommended authentication method and violates least-privilege/security best practices compared with role-based access.

see the dropdown choices in the question first dropdown choice should be later as there is no union in the first choice.
Yes—you’re correct. The dropdown order in the answer explanation appears reversed.
Given the query structure:

kusto 
DeviceLogonEvents 
| extend Table = "DeviceLogonEvents" 
| take 100 
| [Dropdown 1] ( 
    [Dropdown 2] 
    | extend Table = "IdentityLogonEvents" 
    | take 100 
)

1. union — because it follows the first table and precedes the parentheses.
2. IdentityLogonEvents — because it is the table inside the parentheses.

kusto 
DeviceLogonEvents 
| extend Table = "DeviceLogonEvents" 
| take 100 
| union ( 
    IdentityLogonEvents 
    | extend Table = "IdentityLogonEvents" 
    | take 100 
)

but the documentation from the given reference clearly shows security admin can Add/assign initiatives (including regulatory compliance standards)
You’re correct—the answer key and my previous explanation oversimplified this.
In the Microsoft Defender for Cloud permissions documentation, Security Admin can:

• Add or assign security initiatives, including regulatory compliance standards
• Manage Defender for Cloud security policies

• Assign initiatives — Security Admin
• Apply security recommendations — Contributor

need support for ITIL bridge exam

Question 2:
Correct answer: Read-access geo-redundant storage (RA-GRS).
Why:

• Geo-redundant storage (GRS) replicates data to a secondary Azure region, but applications normally cannot read from that secondary region.
• RA-GRS provides the same geo-replication plus read-only access to the secondary region. This directly satisfies all three requirements:

• Zone-redundant storage (ZRS) replicates within availability zones in one region, not across geographic regions.
• Locally redundant storage (LRS) keeps copies within a single datacenter or region and does not provide geographic redundancy.

Question 5:
Correct answer: A — True
Snowflake supports bulk unloading with COPY INTO <location> using a SELECT statement as the source. This allows you to export either:

• All or selected columns from a table
• Filtered rows using WHERE
• Transformed or joined data

sql 
COPY INTO @my_stage/export/ 
FROM ( 
  SELECT customer_id, order_date 
  FROM orders 
  WHERE order_date >= '2025-01-01' 
);

Question 1:
The correct answers are C and D:

• C. Generate a summary of key insights from your data

• D. Build a pivot table based on your data

• A. Customize conditional formatting rules — Excel already supports this directly, and Copilot is not primarily tested as a tool for creating highly specific custom rules.
• B. Insert a custom chart with specific formatting — Copilot can suggest or create visualizations, but detailed, exact chart formatting is generally still a manual Excel task.

Question 60:
Correct answer: C — Information barriers
Microsoft Purview Information Barriers let an organization restrict communication and collaboration between defined groups, such as two departments. They can prevent users in one department from communicating with or sharing information with users in another department—for example, to meet compliance or ethical-wall requirements.
Why the others are incorrect:

• Sensitivity label policies classify and protect content, such as documents and emails; they do not primarily block communication between departments.
• Customer Lockbox controls Microsoft access to customer content during support operations.
• Privileged Access Management (PAM) provides time-limited access to privileged tasks, not department-to-department communication restrictions.

Question 24:
Question 24 correct answer: B — To allow communication between the VPC and the internet.
An Internet Gateway (IGW) is a horizontally scaled, redundant VPC component that enables communication between resources in a VPC and the public internet.
For internet access to work, the following are typically needed:

• An Internet Gateway attached to the VPC
• A route in a subnet’s route table pointing internet-bound traffic to the IGW
• A public IPv4 address or equivalent public connectivity for the resource
• Appropriate security group and network ACL rules

• A: Site-to-site VPN connections use a virtual private gateway or transit gateway, not an IGW.
• C: An IGW does not impose bandwidth constraints.
• D: Load balancing internet traffic is handled by services such as Elastic Load Balancing, not the IGW.

Question 43:
Question 43’s answer key appears to be incorrect. The correct answer should be: Yes.

• Azure App Service is a PaaS service.
• Azure Storage accounts provide managed storage such as Blob, File, Queue, and Table storage. They are generally considered PaaS, because Microsoft manages the underlying hardware, operating system, and storage infrastructure.
• You manage the data and configuration, not the physical infrastructure or storage servers.

Question 11:
Correct answer: B — Store contact details in DynamoDB and the photo object keys in Amazon S3.
Why:

• DynamoDB is well suited for employee records, such as name, department, phone number, and email.
• High-resolution photos are large binary objects, so they should be stored in Amazon S3.
• The DynamoDB item stores the corresponding S3 object key, such as employees/12345/photo.jpg.
• The application uses AWS APIs to:

• A: Base64 encoding increases data size and is inefficient for storing photos in DynamoDB.
• C: Cognito user pools manage authentication and user identities, not a general employee directory with high-resolution photos.
• D: RDS plus EFS is more operationally complex and is unnecessary for this access pattern.

Question 1:
Correct answer: B — With the mobile user license, set up Explicit Proxy.
The customer wants to replace a third-party web proxy for endpoint internet traffic. Prisma Access Explicit Proxy is designed for this use case: client applications send web traffic to the proxy, and Prisma Access applies security controls such as URL filtering, threat prevention, and logging.
Why the other options are incorrect:

• A: GlobalProtect uses a VPN/tunnel-based approach rather than directly replacing an explicit proxy deployment.
• C: A service connection connects Prisma Access to private enterprise networks or data centers; it is not for securing mobile users’ internet traffic.
• D: A corporate access node is associated with private application access, not proxying mobile-user internet traffic.

Question 6:
Correct answer: A — Move rule 1 to the bottom of the list.
I can’t see the role-mapping exhibit itself, but the key concept is ClearPass role-mapping rule order. Rules are evaluated from top to bottom, and a broad rule placed first can match a certificate before a more specific rule gets a chance to assign the intended role.
Typically:

• Specific certificate conditions should be evaluated first.
• Broad or catch-all conditions should be placed later.
• The default role should handle clients that match none of the intended rules.

• Changing the default role does not correct an earlier match.
• Issuer-CN versus Subject-CN depends on the certificate design and cannot be changed without the exhibit.
• Selecting all matches could assign multiple roles and may create conflicting results.

Question 4:
Correct answer: D — Specify an OCSP responder and set the hostname to localhost.
The question describes creating a ClearPass local CA to issue certificates to mobile clients. Those certificates must also be usable for client authentication, which requires a revocation-checking mechanism.

• OCSP allows ClearPass to check whether an issued client certificate has been revoked.
• For certificates issued by a local ClearPass CA, the OCSP responder is hosted on the ClearPass server itself, so the responder hostname should be localhost.
• This lets ClearPass include the appropriate OCSP information in issued certificates and validate certificate status during authentication.

• External validator for EST: relates to validating EST enrollment requests, not the required certificate-status responder.
• SHA-512: changes the digest algorithm but does not provide revocation checking.
• Microsoft Entra ID registration authority: is unrelated to the described on-premises ClearPass CA design.

so why did u mark it as certified incorrectly
You’re right to challenge that. I made an error by accepting the question bank’s explanation instead of checking whether dashboards support endorsement.
I should have identified that:

• Certified is an endorsement applied to supported Power BI items, such as semantic models and reports.
• A Power BI dashboard cannot be endorsed, so executive approval does not make it “Certified.”
• Entity4 should therefore be No badge / Not applicable, if that option exists.

The second question is "Paying electricity for your datacenter is an example of OpEx". According to your explanation this statement would be correct then, right?
Yes, you’re right. My previous explanation was incorrect.
Paying for electricity used by your datacenter is OpEx because it is an ongoing operating cost, not an upfront investment in a long-term asset.
The technically correct selections should be:

1. No — Pay-as-you-go cloud services are OpEx, not CapEx.
2. Yes — Datacenter electricity is an ongoing operational expense.
3. Yes — Deploying or building your own datacenter requires upfront investment and is CapEx.

Question 1:
Correct answer: B — Clustering keys
Snowflake automatically organizes table data into micro-partitions and uses natural clustering based on how data is loaded. However, for large tables where query performance depends on particular columns, you can define a clustering key.
A clustering key:

• Specifies one or more columns or expressions Snowflake should use when organizing table data.
• Helps Snowflake’s automatic reclustering keep related rows grouped together.
• Can improve pruning for queries that frequently filter or join on those columns.

• Micro-partitions: Snowflake’s storage units; users do not directly define them as an override mechanism.
• Key partitions: Not a Snowflake feature.
• Clustered partitions: Not the name of the customer-defined mechanism.

Question 2:
For Question 2, the intended answer is C and D:

• C. Generate a summary of the key points in your document — clearly supported by Copilot in Word.
• D. Insert a custom watermark that has specific text and formatting — likely intended to represent Copilot-assisted formatting and visual customization.

Question 2:
Question 2 tests what Copilot in Word can do, rather than ordinary Word features.

• C. Generate a summary of the key points in your document — Correct.

• D. Insert a custom watermark — The answer key is questionable.

• A. Insert a table of contents and B. Customize page margins are also long-standing Word features, generally performed through Word’s commands rather than Copilot.

Question 25:
Correct answer: A — Read Table3.
Why:

• Table3 is a managed lakehouse table, so it is exposed through the lakehouse SQL analytics endpoint and can be queried with read-only T-SQL.
• The SQL endpoint is read-only. Therefore, you cannot update Table3 or any other table through it.
• A table created as an external Spark table (Table2) isn’t available for querying through the SQL endpoint in this scenario.
• Although Table1 is a Delta table based on a shortcut, the SQL endpoint does not provide an update operation for shortcut data. In any case, “Update the data in Table1” is invalid because the endpoint is read-only.

Question 17:
Correct answer: D — To accommodate a more complex workload.
Increasing a Virtual Warehouse from X-Small to Medium is vertical scaling. The larger warehouse provides more compute resources, which can improve performance for:

• Complex queries
• Large scans or joins
• Resource-intensive transformations
• Queries involving substantial processing

• More users / more queries: This usually calls for multi-cluster warehouses, which add clusters to handle concurrency.
• Fluctuations in workload: This is typically addressed with auto-suspend/auto-resume or multi-cluster scaling, rather than simply choosing a larger warehouse.

Question 1:
Correct answer: Buffer overflow (C).
The program copies a user-controlled string into buffer1, which can store only 10 bytes. If the input exceeds that capacity, the extra data may overwrite adjacent memory, such as other variables, control data, or the return address.
That condition is a buffer overflow vulnerability. It can cause:

• Program crashes or denial of service
• Corruption of data
• Potential execution of attacker-controlled code

• SQL injection: Requires unsanitized input being used in a database query.
• Denial-of-Service: Describes an impact, not the specific vulnerability shown.
• Cross-site scripting: Involves injecting scripts into web pages viewed by other users.

Question 5:
The correct answer is A: Prepare the training data.
Even when the data is consistent and uniform, it must be made ready for machine learning. This usually includes:

• Verifying the data format and schema
• Selecting relevant features and the target column
• Handling missing or invalid values, if present
• Splitting the data into training and evaluation sets
• Registering or making the data available as an Azure Machine Learning data asset

1. Prepare the training data
2. Train the model
3. Evaluate the model
4. Tune hyperparameters, if needed
5. Deploy the model

Question 1:
The answer key shows C, but I believe the best answer is A: Provides a scalable platform for developing and deploying generative AI solutions.
Why A is correct:

• Microsoft Foundry is an enterprise platform aimed mainly at developers, data scientists, and IT teams.
• It supports the full lifecycle: selecting models, connecting enterprise data, evaluating solutions, deploying applications, and monitoring them.
• It is designed to scale a generative AI solution from experimentation to production.

• B: You still need to select and configure an appropriate model; Foundry provides model choices rather than eliminating that requirement.
• C: Business users can use some visual tools, but they are not the primary target audience for building complete solutions.
• D: Foundry may include low-code or visual experiences, but it is not primarily a low-code platform.

Question 1:
Correct answers: C and D

• C. Generate a summary of key insights from your data

• D. Build a pivot table based on your data

• A. Customize conditional formatting rules — Excel already provides conditional-formatting tools, but highly specific rule customization is generally a standard Excel task rather than a core Copilot capability tested here.
• B. Insert a custom chart with specific formatting — Copilot can assist with charts and visualizations, but precise, custom formatting is normally completed manually in Excel.

Question 2:
Answer: B — Add-AzVhd
Add-AzVhd uploads a local, generalized .vhd file to an Azure Storage account as a fixed VHD. This was the traditional process for making an on-premises Hyper-V image available in Azure.
Why the other options are not correct:

• Add-AzVM creates or configures a virtual machine; it does not upload a VHD.
• Add-AzImage creates an Azure VM image resource from an existing managed disk or snapshot. It does not upload the local VHD itself.
• Add-AzImageDataDisk is used for adding data disks to an image, not for uploading the operating-system VHD.

1. Generalize the VM with Sysprep.
2. Upload the VHD using Add-AzVhd.
3. Create an Azure image from that uploaded VHD.

Question 1:
Correct answer: A — Configure a SetupComplete.cmd file in %windir%\setup\scripts.
SetupComplete.cmd runs automatically near the end of Windows Setup, after the operating system has been installed. It is suitable for running initial configuration scripts on newly deployed VMs.
The batch file can invoke your PowerShell scripts, for example:

cmd 
powershell.exe -ExecutionPolicy Bypass -File C:\Scripts\ConfigureVM.ps1

• Logon GPO: Runs when a user logs on, so it is not guaranteed to perform initial VM configuration before use.
• Startup GPO: Runs during startup and may run repeatedly. It also depends on the VM being able to contact the domain.
• Place scripts in a VHD: Merely storing scripts in a disk does not execute them automatically.

Question 5:
Correct answer: A — The steps
In Value Stream Mapping, you first identify the sequence of steps that work passes through—from the initial request or idea to delivery of value. This establishes the workflow being analyzed.
Once the steps are mapped, the team can then identify:

• People and roles involved
• Wait time and processing time
• Bottlenecks and delays
• Active Time, Flow Time, and % Complete and Accurate (%C&A)

• People who supervise the work: roles are relevant, but they do not define the value flow.
• Bottlenecks: these are discovered by analyzing the mapped steps and flow data.
• Active Time, Flow Time, and %C&A: these metrics are collected for the steps after the workflow is understood.

price
I don’t have current pricing information. The full version provides a larger question bank with verified answers. I can also explain any existing Microsoft Fabric exam question or concept from this page.