I work with teams in the US, UK, Europe and the Gulf whose product slows or fails at peak traffic. Reliability you haven’t tested is a guess, so I start with load tests, failover tests and SLOs. They show where your system breaks and which slow requests your users notice first. I fix those before anything else.
Pick the service below that matches your problem, or book a free 30-minute call and I’ll tell you which one fits.
Reliability
Who it’s for. You have a launch or a seasonal peak coming, and your team can’t say where the system breaks. Maybe you already know, because it broke last time. Your alerts wake people at 3 a.m. for noise and stay quiet during the incidents that cost you customers.
What you get.
- A k6 load test suite modeled on your real traffic, with a capacity report showing where the system tops out and why.
- A p99 latency budget for your busiest requests, split across the services each request calls.
- Failover test results. I kill pods, nodes or a database primary on purpose, in staging first, and time the recovery.
- SLOs and an error budget policy that your engineers and your product lead both sign off on.
- An alert review that removes the noise and links each remaining alert to a runbook.
- Dashboards and infrastructure fixes wherever the tests point, on the stack you already run.
How it runs. It starts with a two-week assessment: load tests, a failover test in staging and a review of your alerts. You get a written report that ranks the problems by how soon your users would hit them. I can fix them as a fixed-scope project, or your team can work from the report.
Proof. As part-time senior SRE at Niche Academy, I cut unplanned outages by 90%. I also removed the bottlenecks that capped the requests per second its infrastructure could serve. At Systems Limited, I ran production Kubernetes for Telenor with Prometheus, Grafana, custom exporters and chaos experiments. For Regeneron, I built an observability framework on EKS with the Prometheus stack.
The reliability cookbook shows my method in public. Each recipe pairs a fix for nginx, Linux, Node.js, PostgreSQL or Redis with the test that checks it on your own system.
Read first. I wrote about finding your p99 latency limits before launch day and about starting with SLOs before you hire an SRE.
Fractional CTO
Who it’s for. You run a company with a capable engineering team and no senior technical leader above it. Your CTO may have left, or you aren’t ready to hire a full-time one. Engineers settle architecture calls in pull requests, and nobody owns the trade-off.
What you get.
- A written assessment of your architecture and delivery process, with the risks ranked by what they could cost you.
- Architecture calls made and written down, so your team can read the reasoning months later.
- An infrastructure roadmap tied to your product plans and your next fundraise.
- Help with vendor choices and with hiring engineers, including interviews.
- Part-time senior leadership, on a schedule we agree on at the start.
How it runs. A monthly retainer of one to two days a week, with a three-month minimum. Architecture and hiring decisions take a few months to show up in how your team ships.
Proof. I’m fractional CTO for two companies today. Both engagements are private, so I don’t name them here. I’ve mentored more than 20 engineers. I also founded upzero.io and md0.io and run both.
Read first. I broke down what a fractional CTO costs and when to hire one, so you can judge whether the model suits your stage.
Cloud architecture
Who it’s for. You’re about to make a big infrastructure change, such as moving to Kubernetes, and you want a second opinion first. A large customer may want their data kept in their own country, and your cloud provider has no region there. Or you want a smaller cloud bill without giving up uptime.
What you get.
- An architecture review of your current setup, with findings ranked by risk and by cost.
- Kubernetes platforms on EKS, AKS, GKE or OKE, built for the cloud you already use.
- Terraform for your infrastructure and deploys, written so your team can own it afterward.
- Multi-cloud or multi-region designs that meet regional compliance rules.
- Cloud cost cuts, each one checked against your latency and uptime targets before it ships.
How it runs. A fixed-scope review of one to two weeks that ends in a written report. If you want me to build what it recommends, I quote that as a separate project.
Proof. I’ve spent eight years in cloud infrastructure on AWS, Azure, GCP and Oracle Cloud. At Systems Limited I built a container platform for Wolters Kluwer that deployed Kubernetes clusters on AWS and Azure. For Dubai Municipality, I built a Kubernetes platform for Mendix apps. I hold the AWS Certified Solutions Architect, Professional certification.
Read first. My post on EKS versus running Kubernetes yourself shows how I weigh a managed service against owning the control plane.
What people I've worked with say
12 recommendations on LinkedIn (opens in a new tab)Safoor was an energetic and thorough contributor to our project. He stepped in and solved important problems at a critical time for our company. I recommend without reservation both his work ethic and the quality of his work.
He was instrumental in building our end-to-end infrastructure on OCI Kubernetes, making sure it was scalable, stable, and production-ready.
He helped us optimize the application to run Cypress Test properly and helped us setup our nightly test execution.
I have worked with Safoor on multiple projects and always loved the way he owned and approached even the smallest of tasks assigned.
How we start
On a free 30-minute video call, you describe how your product behaves at your busiest hour. If none of the three services fits, I'll say so.
You deal with me from the start, with no account manager in between. The booking page lists what we cover.