FAI Azure Architecture Review

Delivers a structured Well-Architected Framework review that produces scored pillar assessments, severity-tiered findings, actionable remediation tasks, and an executive scorecard. Transforms informal architecture discussions into documented, trackable commitments with clear owners and timelines.

When to Invoke

Workflow

Step 1 — Collect Architecture Evidence

# Export deployed resources for review
az resource list --resource-group $RG_NAME \
  --query "[].{name:name, type:type, location:location}" \
  --output table > architecture-inventory.txt

# Check for exposed public endpoints
az network public-ip list --resource-group $RG_NAME \
  --query "[].{name:name, ip:ipAddress}" --output table

# Check diagnostic settings coverage
az monitor diagnostic-settings list \
  --resource "/subscriptions/$SUB_ID/resourceGroups/$RG_NAME" \
  --output table

# Check for API keys vs Managed Identity
az resource list --resource-group $RG_NAME \
  --resource-type "Microsoft.CognitiveServices/accounts" \
  --query "[].{name:name, disableLocalAuth:properties.disableLocalAuth}" \
  --output table

Step 2 — Pillar Assessment Worksheets

Score each finding area 1-5 (1=critical gap, 5=exemplary).

#### Reliability

#### Security

#### Cost Optimization

#### Performance Efficiency

#### Operational Excellence

#### Responsible AI

Step 3 — Finding Severity Classification

def classify_severity(score: int, pillar: str) -> str:
    """Score 1-2 in security/reliability = Critical; 1-2 elsewhere = High; 3 = Medium; 4-5 = Low."""
    critical_pillars = {"security", "reliability"}
    if score <= 2 and pillar in critical_pillars:
        return "CRITICAL"
    elif score <= 2:
        return "HIGH"
    elif score == 3:
        return "MEDIUM"
    else:
        return "LOW"

Step 4 — Remediation Priority Table

Step 5 — Executive Scorecard

## Architecture Review Scorecard -- {Date}

| Pillar | Score /5 | Trend | Top Finding |
|--------|----------|-------|-------------|
| Reliability | 2.8 | down | No retry policies on AI calls |
| Security | 2.4 | flat | API key in application config |
| Cost Optimization | 2.6 | flat | PTU underutilized at 45% |
| Performance Efficiency | 3.0 | up | P99 latency 8s vs 3s budget |
| Operational Excellence | 2.8 | flat | 3 resources created manually |
| Responsible AI | 2.6 | flat | No groundedness evaluation |

**Overall Score: 2.7/5 -- NEEDS IMPROVEMENT**
**Critical: 2 | High: 3 | Medium: 5 | Low: 4**
**Review Cadence: Monthly until all pillars score >= 4.0**

Review Checklist

# Before the review session
az resource list --resource-group $RG_NAME --output table     # Inventory
az advisor recommendations list --output table                 # Azure Advisor findings
az security assessment list --output table                     # Defender for Cloud

# After the review -- commit the scorecard
git add docs/architecture-review-$(date +%Y-%m-%d).md
git commit -m "docs: architecture review scorecard $(date +%Y-%m)"

WAF Alignment

Compatible Solution Plays

• All plays — applicable as a cross-cutting governance practice
• Play 02 — AI Landing Zone (infrastructure review)
• Play 17 — AI Observability (operational excellence pillar evidence)

Notes

• Score findings against evidence -- not against intent or the team's roadmap
• CRITICAL findings must have a target date <= 1 week; escalate to engineering leadership if blocked
• Repeat the review after each CRITICAL finding is remediated; do not wait for the full monthly cadence
• Use Azure Advisor and Defender for Cloud findings as objective evidence for scoring