Hi, I'm Ruth Mutua
Platform & Cloud Engineer
Kubernetes in production, infrastructure as code, CI/CD, security hardening and incident response across AWS, Azure and GCP. Focused on systems that are secure by default, automated end to end, and easy for the next engineer to pick up.
$ whoami ruth-mutua $ cat profile.yaml role: platform & cloud engineer location: Nairobi, Kenya Β· EAT (UTC+3) clouds: [aws, azure, gcp] editors: [vim, vscode] focus: - kubernetes in production - infrastructure as code - cloud security & compliance - incident response $
The problem, what I did, what went wrong along the way, and the outcome.
Every fix revealed another problem underneath. What it took to get certificates working again, and the playbook I wrote afterwards.
Read case study βRight-sizing, a cold standby for disaster recovery, and clusters that turn themselves off at night, all in Terraform.
Read case study βFrom Docker Compose on VMs to dedicated clusters in eight phases, including the dependency nobody had written down.
Read case study βPreparing for SOC 2, ISO 27001, HIPAA and GDPR without a compliance platform, starting by checking whether our "green" dashboard was true.
Read case study βA gateway that every AI call goes through, so we know who used what, can cap it, and can switch providers without touching the agents.
Read case study βWhat I use day to day to build, ship and operate infrastructure.
What broke, why, and what I'd do differently.
A peer-to-peer agent mesh has no central coordinator, but it still depends on the platform underneath it: a stable seed address, reachable IPs and network policies that allow the right traffic. Why dead peers linger, and how periodic reconciliation fixes it.
Read article βSome pods serve no traffic yet keep something critical alive. How a standby cluster's secrets went three weeks stale, and why "do-nothing" components need to be documented before someone cleans them up.
Read article βA health check existing isn't the same as a person finding out something broke. Kubernetes probes restart pods and reroute traffic, but nobody hears about it unless you alert on it. Heartbeats, missing check-ins and unusual data volumes catch the silent failures.
Read article βA Java Android app with login and facial recognition through AWS Rekognition.
Read article βOpen-source Python framework for distributed AI agents and workflows, published on PyPI. Release engineering, CLI tooling, security hardening (outbound TLS verification) and contributor documentation.
Peer-to-peer membership and messaging library based on the SWIM protocol.
Nairobi, KenyaI'm a platform and cloud engineer. I'm comfortable stepping into an existing system, reading the code, configs and monitoring, and finding the actual root cause rather than patching symptoms.
I document as I go, with runbooks, architecture decisions and postmortems, and I use AI tools daily, mostly in manual mode, reviewing each step and approving every change.
BSc Computer Technology, JKUAT. Outside engineering: aviation, Formula 1, golf and a lot of coffee. More below β
$ cat /etc/ruth.conf The config that makes me, me.
I love everything about flying: general aviation planes lined up on the apron, the sound of an engine run-up, and the moment the wheels leave the runway. If something is taking off, I'm watching.
Aviation also shaped how I engineer. Checklists are runbooks, redundancy isn't optional, and a go-around is just a clean rollback.
βThe only time you have too much fuel is when you're on fire.β
fuel: 100LL Β· favourite_sound: engine_run_upI run membership for a pan-African community getting more girls and women into cockpits, hangars and control towers: welcoming new members, looking after their welfare, and keeping everyone connected.
And whenever the community needs something new online, from a quick update to a full refresh before the Girls Airshow Africa, giaa.africa is my favourite side project.
Visit giaa.africa βI love the thrill of Formula 1: twenty-two cars at over 300 km/h, and races won or lost by tenths of a second.
The nerdy side hooks me even more: telemetry streaming off every car, tyre strategy that's really capacity planning, and a pit wall making live calls from dashboards. It's an incident room at 300 km/h.
undercut = fresh_tyres β pit_lossGolf is my reset button: fresh air, a long walk, and a problem that looks simple and isn't.
It's a non-deterministic system with one tiny target. Wind, slope and grass all change the outcome, so I read greens like latency graphs and accept the occasional retry.
strokes = par + Ξ΅Every good runbook, postmortem and late-night fix I've written started with a cup of coffee. Probably the second one.
uptime(ruth) β coffeeHave a role, a project or a question about infrastructure? I'm happy to hear from you.