Ibrahim
Cisse
Infrastructure Engineer · Platform & SRE
I build infrastructure and then intentionally break it to understand how it fails. 4 years operating production systems in fintech on AWS at 99.9% uptime.
Bare-metal Kubernetes cluster chaos-tested against node failures, memory exhaustion, and pod crashes. One test exposed a 25-hour outage caused by local storage affinity — documented as full SRE-style postmortems.
CI/CD pipeline with a real-time monitoring dashboard that pulls live metrics from the GitHub Actions API — actual success rates, build times, and run history. Not mock data.
Deployed a 2048 game on EKS with Auto Mode for automated node provisioning, EBS-backed persistent storage, ALB ingress, and dynamic scaling. Full Kubernetes resource lifecycle end-to-end.
Event-driven autoscaling using the KEDA HTTP add-on on Amazon EKS — scales to zero under no traffic and back up instantly under load.
Fully automated multi-node Kubernetes cluster provisioned end-to-end with Terraform — from zero to running cluster as pure infrastructure as code.
Full microservices application deployed on Kubernetes with a complete CI/CD pipeline — container builds through to production manifests.
Zero-downtime Kubernetes version upgrade runbook for production clusters — the kind of documentation you actually need during an on-call incident.
Production-grade AWS VPC architecture with subnets, routing tables, NAT gateways, and security groups — built to Well-Architected Framework standards.
Full observability pipeline — Prometheus scraping metrics, Grafana dashboards, and PagerDuty on-call alerting integrated end-to-end.
Bash monitoring script for a 3-tier stack (PostgreSQL → Flask → Nginx) with retry logic, threshold alerts, Slack webhooks, and alert deduplication to prevent notification fatigue.
Automated system health checks for production Linux environments — CPU, memory, disk, and service status in a single pass.
Network diagnostics script thorough enough to be recognised by Datadog engineers. Covers connectivity, DNS resolution, latency, and routing in one executable.
CI/CD pipeline for a Java Gradle application — automated build, test, multi-stage Docker image creation, and push to Docker Hub triggered on every commit to main.
REST API for patient records and appointment scheduling, built to demonstrate full-stack system ownership from an SRE perspective. Designed with observability and SLOs from the ground up — not bolted on. Python → FastAPI → PostgreSQL → Docker → Prometheus.
E-commerce application used as the primary workload for the AWS multi-infrastructure trade-offs PoC — load-tested across ECS Fargate, EKS, EC2, and Lambda with 239K+ requests.