Mariem Ksontini

– Manage multi-region Azure cloud infrastructure for BMW’s supply chain system iJIX, automating deployments via GitOps with Helm, ArgoCD and GitHub Actions across 5 regions and 12 environments – Refactored Terraform code into reusable modules for core infrastructure components, reducing code duplication and simplifying infrastructure provisioning and management. – Led migration of data streaming platform from Confluent Cloud to self-hosted Kafka on Kubernetes using Strimzi, reducing costs by 92% and improving latency by 53%; automated topic migrations with custom Python scripts. – Designed and implemented a unified observability stack (Grafana Alloy, VictoriaMetrics, Loki, Tempo) reducing Azure Log Analytics costs by 88% and reducing MTTR by 60% through distributed tracing and centralized monitoring. – Migrated nine service deployments from Kustomize to Helm, standardizing GitOps workflows and improving maintainability.

– Spearheaded designing and deploying a monitoring system for Keycloak authentication server using Prometheus and Grafana on Azure Kubernetes, tracking authentication metrics and user activity across Value’s applications. – Provisioned Kubernetes infrastructure on Azure using Terraform. – Implemented CI/CD pipelines with GitHub Actions and ArgoCD, reducing deployment time by 70%. – Created three custom Grafana dashboards visualizing authentication patterns, service availability, and system health.

– Extended DRminer (Docker refactoring detection tool) by developing a weighted quality scoring algorithm in Python to quantify the impact of refactoring techniques on quality attributes such as build time, image size, and maintainability. – Analyzed 500+ open-source projects to identify correlations between Docker refactoring practices and quality attributes, determining which techniques improve or degrade project performance. – Refined weights for Dockerfile refactoring patterns based on empirical impact analysis, contributing to academic research on containerization best practices.

– Developed a MERN stack web platform for NaviScan, an AIoT-based road defect detection system with over 90% detection precision. – Containerized the backend using Docker and hosted it on AWS EC2. – Built and deployed a landing page with HTML, CSS, and JavaScript to showcase project features.