Infrastructure and CI/CD from scratch

Problem

Environments were provisioned by hand, which made every new enterprise deployment slow and inconsistent, and the shared CI runners could not sustain the team's pipeline volume. Both problems compounded as headcount and customer count grew.

Constraints

• Production AWS network was exposed in ways that had to be closed without downtime.
• Multiple environments had to be reproducible rather than hand-tuned one-offs.
• VAPT findings had to be resolved across distributed backend services.

What I built

• Terraform IaC from scratch covering ECS, RDS Multi-AZ, EC2, CloudFront, NAT Gateway, and Secrets Manager.
• Hardened the production AWS network with private subnets, jump servers, and IAM policies eliminating public access exposure.
• Reusable multi-environment provisioning workflows wired into CI/CD deployment.
• Autoscaled self-hosted GitLab Runner infrastructure sustaining 10 to 15 parallel pipelines.
• End-to-end testing infrastructure with dedicated CI/CD integration, catching regressions before enterprise releases.

Result

Infrastructure provisioning time dropped by 97% and deployment turnaround improved by 50%. New enterprise environments became a repeatable operation instead of a project, and public access exposure was eliminated from the production network.