- Provision and run containerized workloads on AWS EKS, ECS Fargate and Azure AKS with autoscaling and rolling deployments.
- Cut average deployment time from 90 minutes to under 15 across 6 product teams by rebuilding multi-environment pipelines in GitHub Actions, GitLab CI/CD and Jenkins.
- Manage AWS infrastructure with Terraform, enforcing consistent tagging, network segmentation and least-privilege IAM.
- Built a Prometheus and Grafana observability stack with alerting on CPU, memory, health and request latency, cutting mean time to detect by 50%.
- Standardized multi-stage Docker builds, 60% smaller images, pushed to ECR with vulnerability scanning at every pipeline stage.
- Removed hard-coded credentials from 100% of deployed repositories through secrets management and environment-specific config injection.
- Cut a client’s infrastructure spend from GBP 9,000 to GBP 3,000 a month by moving dev environments on-premises and replacing purchased public IPs with Cloudflare Tunnel behind a VPN.
Open to remote and contract work
I make infrastructure boring.
DevOps engineer in Lahore. Five years in software, four on the platform layer: Kubernetes on EKS and AKS, Terraform and Terragrunt, CI/CD, and the observability that tells you when any of it breaks.
- checkout0m 22s
- terraform plan1m 12s
- build + scan4m 38s
- push to ecr0m 46s
- argocd sync0m 58s
- blue/green rollout6m 18s
How to read this page. Every number comes from production work at Pikessoft, Northbay or DevSmith. The repositories below are the generalized versions, with client details replaced.
01 / Impact
Infrastructure is judged by what it removes.
Slow deploys, silent failures, surprise bills, credentials in git. Each of these was a number I could move.
- Average deployment time across 6 product teams
- 90 → 14minAverage deployment time across 6 product teamsPikessoft · CI/CD
- Mean time to detect after centralized alerting
- −50%Mean time to detect after centralized alertingPikessoft · Observability
- Deployed repositories with hard-coded credentials removed
- 100%Deployed repositories with hard-coded credentials removedPikessoft · Secrets
- Client infrastructure spend, without losing a dev environment
- £9k → £3k/moClient infrastructure spend, without losing a dev environmentPikessoft · Cost
02 / Selected work
Four platforms, one pattern.
Every project here is a production pattern rebuilt as a reference implementation: provisioned with code, delivered through a pipeline, and proven by what it does under load. Client names and account details are replaced with placeholders.
01Kubernetes platform · GitOpsEKS platformMulti-environment EKS with Terragrunt, Argo CD and blue-green rollouts gated by an automated smoke test.
Stack
Read the repository02Cost-optimized compute · Blue/greenEC2 deployment platformThe low-cost counterpart to a managed orchestrator: same release discipline, no cluster bill.
Stack
Read the repository03ECS Fargate · Terragrunt modulesECS deployment platformA Fargate service with a real data tier, provisioned per environment from the same module set.
Stack
Read the repository04.NET on IIS · Angular on Cloudflare PagesGitLab CI/CD pipelinesTwo production pipelines from one self-hosted GitLab: a Windows backend that backs itself up, and a frontend that ships itself.
Stack
Read the repositoryMore repositories, including the MLOps coursework:github.com/Kamrank47 ↗
03 / Stack
Tools I run, not tools I list.
Grouped by the layer they live in. Everything below has been used to ship or operate something, except where marked as current study.
Cloud & platforms
AWS (EKS, ECS/Fargate, EC2, S3, IAM, RDS, CloudWatch, CodePipeline); Azure (AKS, VMs, App Services, Blob Storage, MSSQL); GCP
Containers
Kubernetes, Amazon EKS, Azure AKS, Docker multi-stage builds, ECR, Kustomize, Helm, Argo CD, Argo Rollouts
Infrastructure as code
Terraform, Terragrunt, AWS CloudFormation, Ansible
CI/CD & source
GitHub Actions, GitLab CI/CD, Jenkins, AWS CodePipeline, Azure DevOps Pipelines, Git, GitHub, GitLab, Bitbucket
Observability
Prometheus, Grafana, Alertmanager, Loki, CloudWatch, Datadog, New Relic
Linux & security
Ubuntu, Amazon Linux, RHEL, Bash, Python, IAM least privilege, secrets management, image scanning, patching, firewall rules, Cloudflare Tunnel, VPN
AI/ML infrastructure
MLOps fundamentals, model serving, containerized GPU workloads. Current study, applied in labs rather than production
04 / Experience
Five years, three teams.
From application code to the platform those applications run on. The path matters: I still read the application logs before I touch the cluster.
- Containerized monolithic applications with Docker for reproducible deployment on AWS EC2.
- Configured EC2 Auto Scaling groups, ALBs, RDS and S3, raising application availability to 99.5% uptime.
- Wrote Bash and Python automation for log rotation, backups, health checks and alerting, returning 8+ hours a week of manual ops work.
- Managed Ubuntu and Amazon Linux servers, including patching, access control and firewall rules.
- Built chat room and ticketing features for a Flutter application.
- Built a Bitcoin wallet application and a WhatsApp media backup application.
Education
B.S. Computer Science
Operating Systems, Computer Networks, Distributed Systems, Cloud Computing
Certification
AWS Solutions Architect (Pasha)
Solutions architecture across compute, networking, storage and IAM.
05 / Writing
Notes from the work.
The problems that took longer than the documentation suggested. Written on Medium, mostly so the next person finds it faster.
06 / Contact
Let’s make your deploys boring.
Open to remote and contract roles. Based in Lahore, PKT (UTC+5), and comfortable writing the runbook as well as the pipeline.