About
I build software and I run the infrastructure it lives on. Since 2009 I've done both professionally — and Linux has been my operating system for as long as I've been using computers at all.
mah3uz
Linux systems & software engineer · since 2009
I’m Mahfuz Shaikh, known online as mah3uz. For more than seventeen years I have worked on both sides of the line that many organisations draw between building software and running it. I write applications, and I administer the Linux servers and infrastructure they depend on — provisioning machines, managing configuration, automating deployments, securing and patching systems, and keeping production healthy long after launch day.
I started out in 2009 as a freelancer on oDesk, now Upwork, and spent five years there taking on development and server work for clients I knew only through the platform. That is where I learned that the job isn’t finished until the client can understand what was done and the next person can maintain it.
Linux has been my operating system since I first started using computers. That shaped how I work: I would rather understand the whole system than one layer of it, and I am most at home in a terminal, reading logs and source code until the behaviour makes sense.
The most expensive problems I have solved lived in the gap between application and infrastructure: the slow page that was really a database configuration, the failed deploy that was really a full disk, the intermittent error that was really a DNS cache, the “random” restart that was really a memory limit. When the same person understands the code and the machine, those gaps close quickly — and, more importantly, the next system gets designed so they don’t open at all.
It also changes how I write software. I build things that are pleasant to operate: they fail loudly and safely, log what an on-call engineer needs, start and stop cleanly, back themselves up, and upgrade without drama. You can see that in the products I build today — a CMS whose upgrades refuse to run over modified files, a system monitor whose first rule is that the last reading before a crash is already on disk, a lock-screen manager designed so it can never lock you out.
I like small, frequent releases, written changelogs and runbooks, and automation that anyone on the team can run. I would rather spend an afternoon making a failure mode safe than a week hoping it doesn’t happen. I keep things simple on purpose, and I take security seriously by default rather than as a phase at the end.
Through my company, mah3uz – IT Consultancy, I take on a limited amount of client work alongside my own products — engineering, infrastructure, or the place where the two meet. If you have a platform that needs to be built carefully and run reliably, tell me about it.
Principles
Most of what goes wrong in production was decided at a desk months earlier. I design software with the operator in mind — clear failure modes, honest logs, safe defaults — because the operator is often me.
Proven technology — Linux, PostgreSQL, SQLite, Rails, plain HTML — leaves room to spend effort where it shows. The craft is in the details people touch every day.
Do it by hand once to understand it. Write it down the second time. Automate it the third. Deploys, backups and server builds should be repeatable by anyone, including future me at 3am.
A release script that checks everything before it does anything public. A daemon that stops on a configuration typo instead of silently disabling something. Loud and safe beats quiet and wrong.
Least privilege, private sockets, signed releases, secrets kept out of logs and history. The secure path should be the path of least resistance.
Changelogs for the people running the system, runbooks for the people fixing it, documentation for the people using it. Software nobody can operate isn't finished.
Timeline
The milestones I can point to. Most client work is private, so the public record starts later than the career does; where there is something to see, the milestone links to it.
Linux has been my operating system since I first sat down at a computer. It shaped everything that followed — a habit of reading the source, living in a terminal, and wanting to understand the whole system rather than one layer of it.
From 2009 to 2014 I worked as an independent contractor on oDesk (Upwork, as it became), on both sides from the start — building web applications for clients, and installing, configuring and maintaining the Linux servers they ran on. Five years of scoping, delivering and supporting work for clients I met only online taught me to communicate plainly and leave every system documented.
My first public repository. Most professional work — for clients and employers — is private, so the public record from here on is a partial one.
After five years of freelancing I founded my own company, mah3uz – IT Consultancy, which I still run today — software engineering and systems work for clients, owning each project from the server to the front end.
Working with Huginn, a self-hosted Ruby system for building agents that watch the web and act on events — an early taste of automation as infrastructure.
From 2016 to 2018 I worked as a system engineer for Jacos, a Japanese software company, on both sides of the work — developing its software, and building, deploying to and maintaining the servers it ran on.
Working with WireGuard and Subspace, a web interface for managing VPN peers — private networking without a third-party service in the middle.
The start of a long engagement with TLC, an Australian lender of medical, dental and lifestyle payment plans — building its CRM in Laravel, designing the API partners use to submit loan applications, and running the servers underneath.
Proxmox VE on Hetzner dedicated servers, self-hosted Matrix with its sticker tooling, and umbrelOS for apps at home.
Building TourHub in Ruby on Rails — verified crew profiles, contracts and payments secured through Stripe — and the staging and production environments it ships through.
A full redesign of tide, a self-hosted video meeting service, with an installable app and a screen-share popout; and a Dux Shell plugin that brings EarPort's AirPods support from GNOME to Hyprland.
A peer-to-peer, end-to-end encrypted messenger in Rust, with a terminal client and an app for Linux, macOS, Android and iOS, all on one shared core — an exercise in applied cryptography and honest engineering.
A schema-driven CMS you install as your own Rails application, and the Rust command-line tool that goes with it — 25 public releases in its first weeks.
A Rust and Qt theme manager for the Linux login and lock screens, built around safety — a lock that can't lock you out, a root helper that trusts nothing.
A SQL client that lives in the terminal, for PostgreSQL, MySQL, MariaDB and SQLite — the database tool I wanted for my own work, built to production standards.
A terminal multiplexer with a native window, built for running many coding agents at once — a daemon that owns every program, and a window that always shows which one needs you.
A crash-proof system monitor for Linux, whose first rule is that the last reading before a crash is already on disk.
Seventeen years in, the curiosity hasn't worn off. I'm still trying new languages, new hardware and new ways to run a server, mostly in the homelab first, and building the tools I wish existed. This site is one of them, on Nibble, deployed with Kamal to a server I run.
Contact
Tell me what you’re working on. A product that needs a steady pair of hands, a question about one of my tools, or just a hello: it all lands in the same inbox.
