I keep web infrastructure online and engineers unblocked.
I'm a Technical Support Engineer who lives in the terminal.
I troubleshoot production web hosting — when a site is down, slow, or throwing 500s, I'm the person who SSHes in, reads the logs, finds root cause, and ships a durable fix without taking the rest of the box down with it.
Day to day that means OpenLiteSpeed / Nginx vhosts, PHP-FPM pools, DNS and SSL, MySQL/MariaDB tuning, WordPress incidents, and the occasional 2am OOM-killer mystery. I document everything so the next engineer doesn't have to relearn it.
Sites across a shared VPS intermittently 502'd with DB-connection errors; mysqld was being OOM-killed repeatedly.
One PHP-FPM pool with an oversized min_spare was pinning 50+ idle workers, thrashing swap until the kernel killed the lowest-priority process — mysqld.
Right-sized the pool's pm.* values, graceful-reloaded FPM, and hardened mysqld with OOMScoreAdjust=-600 so it stops being the kernel's first victim.
A Node app behind OpenLiteSpeed kept serving the default self-signed certificate; Let's Encrypt issuance failed.
The catch-all proxy context forwarded /.well-known/acme-challenge/ to the Node app, so certbot's webroot check 404'd.
Added a permanent static carve-out for .well-known in a persistent vhost include, issued an ECDSA cert via certbot --webroot, then attached SSL — survives panel regeneration and auto-renews.
A store went sluggish (not down) under heavy CPU and swap; the site itself looked healthy.
A faceted-navigation scraper hammered uncacheable filter URLs, forcing repeated unserialize of a 67MB layered-nav transient and saturating CPU.
Added a targeted query-string block at the web-server layer (preserving legitimate filtering), flushed the bloated transients, and locked the origin firewall to Cloudflare ranges.
Ahsanullah University Of Science & Technology
Foundations in networking, operating systems, and databases that underpin my infra work today.