AI & Cloud Platform Architect

Enterprise microservices platform with production Kubernetes, Helm charts, multi-cluster deployment across AKS/EKS/GKE, GitHub Actions CI/CD, ArgoCD GitOps, delivering 99.9% uptime with containerized file management APIs.

apiVersion: v1
kind: Deployment
metadata:
  name: camp-backend
spec:
  replicas: 3
  strategy:
    type: RollingUpdate
  template:
    spec:
      containers:
      - name: backend
        image: ghcr.io/elishatheodore/camp:latest

Production-grade MLOps for credit risk & fraud detection with MLflow experiment tracking, model registry, FastAPI serving, drift monitoring via Evidently AI, automated retraining, and 95%+ accuracy gates on AKS.

@app.post("/predict")
async def predict(request: TransactionRequest):
    features = feature_engineering(request)
    risk_score = model.predict(features)
    return {
        "risk_score": risk_score,
        "model_version": "v2.1.0",
        "timestamp": datetime.now()
    }

Cloud-agnostic Infrastructure as Code supporting Azure, AWS, and GCP with modular Terraform design, environment-based deployments, multi-cloud CI/CD, security scanning, and drift detection via GitHub Actions.

resource "azurerm_virtual_network" "main" {
  name                = var.vnet_name
  address_space       = var.vnet_address_space
  location            = azurerm_resource_group.main.location
  resource_group_name = azurerm_resource_group.main.name
}

Production ETL pipeline following Medallion Architecture (Bronze/Silver/Gold) on Brazilian e-commerce dataset with 99,441 orders, DuckDB, Pandas, Pytest, and GitHub Actions, delivering R$ 13.5M+ revenue analytics.

def create_fact_orders():
    sql = """
    SELECT order_id, total_revenue,
      delivery_days, delivery_status,
      sentiment, product_category
    FROM silver.orders
    WHERE status = 'delivered'
    """
    return duckdb.sql(sql).to_df()