DevOps, CI/CD & Deployment
Ship on every merge, sleep through the night, recover in minutes.
A feature is not done when it merges — it is done when it is running in production, monitored, backed up, and deployable again thirty seconds after the next change. That last mile is where I bring DevOps discipline to small teams that cannot justify a dedicated platform engineer.
My day-to-day is Vercel for Next.js products, AWS for services that need more control, and GitHub Actions for CI. Domains, SSL, redirects and environment promotion configured so that staging honestly mirrors production — because most 'works on my machine' bugs are really configuration bugs.
I also do rescue work: sites with broken builds, expired certificates, misconfigured redirects or databases that nobody dares touch. Finding the root cause, fixing it, and documenting how to prevent it.
What you get
- CI/CD pipelines with GitHub Actions — test, build, deploy
- Vercel projects: domains, SSL, redirects, preview deployments
- AWS setup: EC2, S3, Route 53, IAM with least-privilege policies
- Environment and secrets management across dev/staging/prod
- Uptime monitoring and error alerting
- Database backups and a restore procedure someone has actually tested
Tools I work with
How we work
Baseline
We document what is running, where, and how it deploys today — including the parts only one person knows.
Automate
The deploy path becomes a pipeline: every push builds, tests and ships to a preview, and merges reach production without human steps.
Observe
Monitoring and alerts on the things that matter — errors, latency, disk, certificate expiry — so you hear about problems before users report them.
Recover
Backups with a tested restore, and a short written runbook. Recovery you have never rehearsed is not recovery, it is hope.
Common questions
Vercel or AWS for a Next.js site?
Vercel for most products — it runs Next.js natively and removes a class of infrastructure problems. AWS earns its complexity when you need long-running processes, specific compliance, or heavy non-HTTP workloads.
My site shows a certificate error — can you fix it?
Yes. Common causes are a domain not fully assigned to the host, a missing www/apex redirect, or DNS pointing somewhere stale. I diagnose the actual certificate being served and fix the configuration, then verify.
Do you offer ongoing maintenance?
Yes — a monthly arrangement covering dependency updates, monitoring response, small fixes and a deploy window. Bigger feature work gets scoped separately.