Documentation

Quick Start Guide

Get perfecX Comply up and running in your organization in under 10 minutes.

Prerequisites

Python 3.8+ or Node.js 16+
Administrative access to your AI/ML platforms
A perfecX Comply API key (sign up at perfecxion.ai)
Basic understanding of your compliance requirements (EU AI Act, NIST AI RMF, SOC 2, etc.)

Step 1: Install the SDK

Python

pip install perfecxion-comply

Node.js

npm install @perfecxion/comply

Step 2: Initialize perfecX Comply

Python Example

from perfecxion_comply import ComplianceClient

# Initialize the compliance client
client = ComplianceClient(
    api_key="your-api-key",
    organization_id="your-org-id"
)

# Configure compliance frameworks
client.configure_frameworks([
    "EU_AI_ACT",
    "NIST_AI_RMF",
    "SOC_2_TYPE_II",
    "ISO_42001"
])

# Set up automated scanning
scan_config = {
    "scan_interval": "daily",
    "include_models": ["production/*"],
    "risk_threshold": "medium",
    "auto_remediate": True
}

client.setup_automated_scanning(scan_config)

Node.js Example

import { ComplianceClient } from '@perfecxion/comply';

// Initialize the compliance client
const client = new ComplianceClient({
    apiKey: 'your-api-key',
    organizationId: 'your-org-id'
});

// Configure compliance frameworks
await client.configureFrameworks([
    'EU_AI_ACT',
    'NIST_AI_RMF',
    'SOC_2_TYPE_II',
    'ISO_42001'
]);

// Set up automated scanning
const scanConfig = {
    scanInterval: 'daily',
    includeModels: ['production/*'],
    riskThreshold: 'medium',
    autoRemediate: true
};

await client.setupAutomatedScanning(scanConfig);

Step 3: Register Your AI Models

# Register an AI model for compliance tracking
model = client.register_model({
    "model_id": "customer-churn-predictor",
    "name": "Customer Churn Prediction Model",
    "version": "2.1.0",
    "type": "classification",
    "purpose": "Predict customer churn probability",
    "data_categories": ["customer_behavior", "transaction_history"],
    "deployment_env": "production",
    "risk_category": "medium",
    "metadata": {
        "team": "data-science",
        "framework": "tensorflow",
        "training_date": "2024-01-10"
    }
})

# Perform initial compliance assessment
assessment = client.assess_model(model.id)
print(f"Compliance Score: {assessment.overall_score}/100")
print(f"Violations Found: {len(assessment.violations)}")
print(f"Risk Level: {assessment.risk_level}")

Step 4: Configure Compliance Policies

# Define compliance policies
policies = client.create_policy_set({
    "name": "Production AI Governance",
    "rules": [
        {
            "type": "bias_detection",
            "threshold": 0.05,
            "protected_attributes": ["gender", "race", "age"],
            "action": "alert_and_block"
        },
        {
            "type": "data_privacy",
            "require_anonymization": True,
            "pii_categories": ["email", "phone", "ssn"],
            "retention_days": 90
        },
        {
            "type": "model_drift",
            "drift_threshold": 0.1,
            "check_frequency": "weekly",
            "auto_retrain": True
        },
        {
            "type": "documentation",
            "required_docs": ["model_card", "risk_assessment", "bias_report"],
            "update_frequency": "monthly"
        }
    ]
})

# Apply policies to models
client.apply_policies(
    policy_set_id=policies.id,
    model_filters={"deployment_env": "production"}
)

Step 5: Monitor Compliance Status

Once configured, perfecX Comply provides real-time monitoring through our dashboard:

Compliance Dashboard

Real-time compliance status across all frameworks

Risk Assessment

Automated risk scoring and mitigation recommendations

Audit Reports

Automated generation of compliance documentation

Violation Alerts

Instant notifications for compliance violations

# Generate compliance report
report = client.generate_compliance_report(
    framework="EU_AI_ACT",
    format="pdf",
    include_evidence=True
)

print(f"Report generated: {report.url}")
print(f"Compliance Status: {report.status}")
print(f"Next Audit Date: {report.next_audit_date}")

# Get real-time metrics
metrics = client.get_compliance_metrics()
print(f"Overall Compliance: {metrics.overall_score}%")
print(f"Models in Compliance: {metrics.compliant_models}/{metrics.total_models}")
print(f"Active Violations: {metrics.active_violations}")

MLOps Integration Example

Integrate perfecX Comply with your existing MLOps pipeline:

# Example: MLflow Integration
from mlflow import log_metric, log_param
from perfecxion_comply import MLflowIntegration

# Initialize MLflow integration
px_mlflow = MLflowIntegration(client)

# Automatic compliance checks during model training
with mlflow.start_run():
    # Train your model
    model = train_model(X_train, y_train)
    
    # Automatic bias detection
    bias_report = px_mlflow.check_bias(
        model=model,
        test_data=X_test,
        protected_attrs=['gender', 'race']
    )
    
    # Log compliance metrics
    log_metric("bias_score", bias_report.score)
    log_metric("fairness_score", bias_report.fairness)
    log_param("compliance_framework", "EU_AI_ACT")
    
    # Register model with compliance metadata
    px_mlflow.register_compliant_model(
        model=model,
        model_name="customer_churn_v2",
        compliance_report=bias_report
    )

Next Steps

Read the full installation guide Explore the API reference View compliance framework guides

Need Help?

Our compliance experts are here to help you navigate AI governance requirements.

Contact Support Browse Documentation