I'm Arslan Asad Qazi, and I build, secure, and deploy the systems that get code from a commit to production without breaking or leaking anything along the way.

Final-year Computer Science student at FAST-NUCES Karachi, graduating in June 2027. My focus is DevOps, Cloud, and DevSecOps engineering: the practical side of getting software built, tested, secured, and deployed without someone babysitting every step.

The parts that get skipped are what I care about most. Pipelines that break quietly, infrastructure that isn't built to scale, security checks added too late instead of built in from the start. Fixing that means designing systems that hold up on their own, from the first commit to the last deploy.

A design background came first, and it still shapes how I build, not just how something runs, but how it feels to use. That mix shows up in the details: clean, secure, and made to last.

Always learning the next layer of the stack. Always looking for the next hard problem worth solving.

This originally ran on a self-hosted pipeline: every push built a Docker image, scanned it with Trivy, pushed it to GHCR, then an Ansible playbook SSHed into a single Oracle Cloud VM to pull and recreate the container, with Caddy handling the reverse proxy and HTTPS. That setup worked well for weeks.

It moved to Vercel after Oracle's free tier locked the account over a routine card verification charge, with no reliable way to recover access in a reasonable time. For a portfolio site that needs to actually stay up, that's not an acceptable failure mode, so the trade-off flipped: managing the VM myself stopped being worth the reliability risk.

The self-hosted path isn't gone, it's just not what serves this site right now. The Dockerfile, Ansible playbook, and Caddy config are still in the repo and still tested in CI (build, lint, Trivy scan on every push), they just don't deploy anywhere at the moment.

The full reasoning is written up as ADRs in docs/adr/ in the repo.