#Projects
Selected work that shows how I think about reliability, automation, and real-world impact.
Teleport Cluster Upgrade Automation
DevOps Engineer • 2025
Designed and implemented an automated process to upgrade 200+ Teleport-connected nodes from v11 to v17 using the internal Teleport network.
- →Wrote robust Bash scripts to download, verify, and replace Teleport binaries on remote nodes.
- →Used systemd units and careful rollout strategy to minimize downtime.
- →Documented a safe upgrade path compatible with different Teleport versions across clusters.
AFMS Infrastructure Automation & Monitoring
DevOps Engineer • 2024 – 2025
Automated CI/CD, containerization, and observability for AFMS Stations/Logistics deployments across dozens of fuel stations.
- →Built GitLab pipelines to build and push images like petrosolutions/api, collector, synchronizer, and AFMS station images.
- →Implemented Docker Swarm deployment configs using secrets for DB credentials and sensitive tokens.
- →Created Prometheus scrape configs and Grafana dashboards to monitor availability, SSL expiry, and performance.
Reverse SSH for LAN-Only Fuel Stations
DevOps Engineer • 2024
Implemented a reliable reverse SSH mechanism for stations with no public IP access, enabling remote support and automation.
- →Generated root SSH keys and configured authorized_keys on a central VPS.
- →Wrote systemd services to keep tunnels alive and reconnect on failures.
- →Integrated with Teleport for secure, audited access.
Debian/Ubuntu Mass Provisioning & Imaging
DevOps Engineer • 2024
Created a repeatable blueprint image for industrial stations and automated its distribution to 100+ machines.
- →Used LVM snapshots and partclone to create minimal, consistent disk images.
- →Automated hostname, network, and Docker stack customization with Ansible.
- →Reduced setup time per machine from hours to minutes.
Medical Analysis AI Web App
Side Project / Founder • Ongoing
Designing an app that reads lab analysis results, visualizes them, and explains them in Moroccan Darija using text-to-speech.
- →Model to interpret reference ranges and flag abnormal values.
- →Visualization of lab results so patients can understand their status.
- →Darija TTS output to make medical terminology accessible to non-experts.