DevOps & Cloud Engineer
Mubashir
DevOps and cloud engineer at Kiebot. Runs the CI/CD, Kubernetes, and observability stack that backs Kiebot’s client deployments.
About Mubashir
Mubashir is the DevOps and cloud engineer who keeps Kiebot’s production environments boring — in the best possible way. He builds the CI/CD pipelines, Kubernetes platforms, and observability stacks that let Kiebot’s engineering teams ship to production confidently and safely. His work spans infrastructure-as-code, secrets management, cost optimisation, and incident response. He writes about pragmatic DevOps, how to build a real platform team, and what production reliability looks like for small engineering orgs.
Recent highlights
- 1
Built Kiebot’s standard GitOps deployment pipeline used across client projects.
- 2
Owns the on-call rotation and incident-response playbook.
- 3
Cut average client cloud spend by 30% through targeted optimisation reviews.
Articles by Mubashir
GitOps for Small Teams: Production-Grade Deployments Without a Platform Team
You do not need a five-person platform team to run GitOps. Here is the minimal, production-grade setup we use at Kiebot for client projects.
Read article →
LangChain in Production: What Works and What Hurts
LangChain gets you to prototype fast. Then production hits you in the face. Here is what we keep, what we replace, and why.
Read article →
Small LLMs Are Eating the Edge
Three to seven billion parameter models now handle a surprising amount of real work. Here is where small wins and where it does not.
Read article →Questions Mubashir gets asked
What is Mubashir’s first DevOps fix for a struggling startup?
A reliable, reproducible deployment pipeline. Nothing slows a startup down more than scary, manual deployments. Fix that first and most other engineering problems get cheaper.
Kubernetes or managed PaaS for an early-stage team?
Start with a managed PaaS unless you already have someone on the team who lives and breathes Kubernetes. Adopt Kubernetes when the cost and rigidity of the PaaS starts hurting more than the complexity of running your own platform.