I'm Ashiwani, a Linux DevOps and SRE engineer in Abu Dhabi. For 7+ years I've looked after systems that can't go down. Right now that's the infrastructure behind five UAE airports, running on Kubernetes, OpenShift and AWS at 99.9% uptime.
➜ ~ kubectl get nodes -o wide
✔24/24 nodes Ready · 5 clusters · 3 regions
➜ ~ uptime --check --env production
✔99.9% over the last 12 months
➜ ~ whoami
ashiwani — Linux DevOps / SRE, Abu Dhabi
I've spent seven years on systems people only notice when they break: airport infrastructure, hospital platforms, telecom networks. Right now I work on the UAE ICP programme, keeping five UAE airports at 99.9% uptime for more than 50 million passengers a year. I'm originally from India and now live in Abu Dhabi.
Almost everything I run is open source: Linux, Kubernetes, Terraform, Ansible. It's what I learned on, and it's what I trust in production.
Infrastructure for five UAE airports on the ICP programme, including Abu Dhabi and Sharjah, plus Muscat International in Oman. There's never a good time for an airport to go offline.
Cloud migrations, pipelines, on-call firefighting. The unglamorous work that keeps your product shipping.
Infrastructure doesn't screenshot well, so I've drawn what each project actually looked like.
Broke monolithic apps into services and moved them onto Kubernetes, with Argo CD deploying straight from Git. Deployments got about 40% faster.
Jenkins and GitLab CI pipelines that build, test and run security scans on every change. Deployments that used to take hours now take minutes.
Wrote Terraform for our AWS setup, from VPCs to databases. A new environment went from days of manual work to a few minutes.
Set up SLOs, Prometheus and Grafana dashboards and alerting that catches problems before users do. Uptime has held at 99.9%.
I also build web apps end to end: Next.js and React on the front, Node.js and a real database behind it, shipped through CI/CD. This site is one of them.
Rolled out Azure DevOps across the whole delivery flow: Boards for planning, Repos with branch policies, multi-stage YAML pipelines and gated releases to production.
Seven years in aviation, healthcare and telecom. When things break there, people notice fast.
Astek Middle East
Linux and DevOps for the UAE ICP programme, placed with Astek's client Idemia. I look after the systems at 5 UAE airports that handle 50M+ passengers a year, kept at 99.9% uptime. Incident response got 70% faster once we brought in DevSecOps.
NMC Healthcare
Ran 400+ servers behind the hospital EMR systems at 99.8% availability, and cut AWS and Azure costs by 35%.
Astek Middle East · Muscat International Airport
On-site at Muscat International Airport, taking the aviation programme to Oman. I'm still on this project and support it alongside the UAE airports.
CoreHive Computing
Looked after 400+ SUSE Linux servers for telecom clients. Automated provisioning with Ansible, which cut setup time by 60%.
Dilip Buildcon
Ran the company's Linux and VMware environments and wrote the automation scripts that made day-to-day operations about 40% more efficient.
Microsoft
Managing Azure identities, governance, storage, compute, virtual networking, and monitoring
Microsoft
Designing and implementing DevOps processes, CI/CD, and release strategies on Azure
Amazon Web Services
Cloud architecture and design certification
Red Hat
Red Hat Certified System Administrator & Red Hat Certified Engineer
Cisco
Cisco Certified Network Associate
Juniper Networks
Juniper Networks Certified Associate - Cloud, Security & Junos
Rajiv Gandhi Proudyogiki Vishwavidyalaya, Madhya Pradesh - First Class
Haryana State Board of Technical Education, Haryana
The most interesting problems right now sit at the intersection of reliability engineering and machine learning. Here's where I'm looking.
Using ML to spot trouble early and scale ahead of load, instead of paging someone after it has already broken.
Most monitoring still just tells you something broke. I'm interested in systems that can fix the common failures on their own.
Models should go through the same process as code: versioned, tested and shipped through a pipeline.
Kubernetes suits AI workloads well. The goal is to make serving a model as routine as deploying any other service.
Tell me what you're building and where it hurts. I'll tell you honestly whether I can help. I usually reply within a day.
Abu Dhabi, UAEPhone: +971 566182303ashvanikumar109@gmail.com
