AI-102 Certification Prep: Part 1 – Planning & Implementing Azure AI Solutions

Welcome to Part 1 of the AI-102 Certification Series, designed to help you master the skills needed to become an Azure AI Engineer Associate. This series aligns with Microsoft’s official learning paths, and in this instalment, we’ll focus on Learning Path 1: Get Started with Azure AI Services. By the end of this guide, you’ll build a real-world AI solution while covering all six modules of the learning path. Let’s dive in!

About the AI-102 Certification Series

The AI-102 exam tests your ability to design, implement, and manage AI solutions on Azure. This 6-part series mirrors the six official learning paths, breaking down complex topics into actionable steps:

Part 1: Get Started with Azure AI Services (This Blog)
Part 2: Create computer vision solutions with Azure AI Vision
Part 3: Develop natural language processing solutions with Azure AI Services
Part 4: Implement knowledge mining with Azure AI Search
Part 5: Develop solutions with Azure AI Document Intelligence
Part 6: Develop Generative AI solutions with Azure OpenAI Service

Stay tuned for upcoming parts!

Plan and Prepare AI Solutions

Key Skills:

• Define use cases (e.g., sentiment analysis, fraud detection).
• Choose Azure services (e.g., Text Analytics, Computer Vision).
• Design data pipelines (storage, processing, deployment).

Azure Portal Tip: Use the Azure Architecture Center for reference designs.

Create and Consume AI Services

Key Skills:

• Deploy Cognitive Services (e.g., Text Analytics).
• Call APIs using SDKs (Python, C#, etc.).
• Store results in databases like Cosmos DB.

Azure Portal Tip: Test APIs instantly using the Cognitive Services “Quickstart” tab.

Secure AI Services

Key Skills:

• Store secrets in Azure Key Vault.
• Restrict network access with Private Link.
• Enable encryption for data at rest.

Azure Portal Tip: Use Azure Policy to enforce compliance.

Monitor AI Services

Key Skills:

• Track latency and errors with Azure Monitor.
• Query logs using Log Analytics.
• Set up alerts for SLA breaches.

Azure Portal Tip: Pin key metrics to your dashboard for real-time visibility.

Deploy AI Services in Containers

Key Skills:

• Containerize apps with Docker.
• Deploy to Azure Kubernetes Service (AKS).
• Auto-scale workloads.

Azure Portal Tip: Use Azure Container Registry (ACR) to manage Docker images.

Use AI Responsibly

Key Skills:

• Detect harmful content with Azure AI Content Safety.
• Audit models for fairness and bias.
• Document ethical guidelines.

Azure Portal Tip: Explore the Content Safety Studio to test moderation policies.

Hands-On Example: Building a Customer Feedback Analyser

Let’s build a system that analyzes hotel reviews using Azure Text Analytics while applying all six modules.

Step 1: Plan & Prepare
Use Case: Analyze sentiment and key phrases in reviews.

Architecture:

• Data Source: CSV files in Azure Blob Storage.
• AI Service: Text Analytics for sentiment analysis.
• Database: Cosmos DB for results.

Step 2: Create & Consume

1. Deploy Text Analytics:
2. In the Azure Portal, create a Text Analytics resource.

3. Grab the API key and endpoint from the Keys & Endpoint tab.

4. Analyze Reviews:
   

from azure.ai.textanalytics import TextAnalyticsClient
from azure.core.credentials import AzureKeyCredential

key = "<YOUR_KEY>"
endpoint = "<YOUR_ENDPOINT>"
client = TextAnalyticsClient(endpoint, AzureKeyCredential(key))

reviews = ["The staff was friendly, but the Wi-Fi was slow."]
response = client.analyze_sentiment(reviews)
print(f"Sentiment: {response[0].sentiment}")

Step 3: Secure
Store the API key in Azure Key Vault:

from azure.keyvault.secrets import SecretClient
from azure.identity import DefaultAzureCredential

credential = DefaultAzureCredential()
client = SecretClient(vault_url="<VAULT_URL>", credential=credential)
client.set_secret("TextAnalyticsKey", key)

Step 4: Monitor

• In Azure Portal, go to Text Analytics → Monitoring → Metrics.
• Track Total Calls and Latency.

Step 5: Deploy in Containers
Push your app to Azure Kubernetes Service (AKS):

# Kubernetes Deployment File
apiVersion: apps/v1
kind: Deployment
metadata:
  name: feedback-analyzer
spec:
  replicas: 2
  template:
    spec:
      containers:
      - name: analyzer
        image: myacr.azurecr.io/feedback-analyzer:v1

Step 6: Ensure Responsible AI
Use Azure AI Content Safety to flag harmful reviews:

from azure.ai.contentsafety import ContentSafetyClient

client = ContentSafetyClient(endpoint="<CONTENT_SAFETY_ENDPOINT>", credential=AzureKeyCredential(key))
response = client.analyze_text(text="This is offensive!")
if response.severity > 0:
    print("Content flagged as harmful.")

Conclusion

You’ve now completed Part 1 of the AI-102 series! By building a customer feedback analyzer, you’ve:

• Planned an AI solution.
• Created and secured Azure AI services.
• Monitored performance and deployed in containers.
• Applied responsible AI practices.

Stay tuned for Part 2, where we’ll tackle Computer Vision! For updates, subscribe to this blog or follow me on LinkedIn.

Happy coding, and see you in the next installment! 🚀

👉 Pro Tip: Try replicating this example using your Azure subscription!

---

💬 Let’s Discuss!
Have questions about Learning Path 1? Drop a comment below!