<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Paramjyot Singh Tooray on Medium]]></title>
        <description><![CDATA[Stories by Paramjyot Singh Tooray on Medium]]></description>
        <link>https://medium.com/@param_2ray?source=rss-d9df9d97174a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*dmbNkD5D-u45r44go_cf0g.png</url>
            <title>Stories by Paramjyot Singh Tooray on Medium</title>
            <link>https://medium.com/@param_2ray?source=rss-d9df9d97174a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 23:53:47 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@param_2ray/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Building a Production-Grade Kubernetes Platform on AWS with Amazon EKS]]></title>
            <link>https://medium.com/@param_2ray/building-a-production-style-kubernetes-platform-on-aws-with-amazon-eks-e27d533115ce?source=rss-d9df9d97174a------2</link>
            <guid isPermaLink="false">https://medium.com/p/e27d533115ce</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[terraform]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Paramjyot Singh Tooray]]></dc:creator>
            <pubDate>Thu, 23 Jul 2026 01:42:39 GMT</pubDate>
            <atom:updated>2026-10-27T22:00:09.762Z</atom:updated>
            <content:encoded><![CDATA[<h3>TL;DR</h3><p>I built a Kubernetes platform on AWS using production-grade engineering practices, with Amazon EKS, Terraform, GitHub Actions, ArgoCD, Traefik, cert-manager, ExternalDNS, Prometheus, Grafana, Amazon Route 53 and GitOps.</p><p>The objective wasn’t simply to deploy an application. It was to design and operate a Kubernetes platform using production engineering practices around automation, networking, security, observability and continuous delivery.</p><p>The result is a platform that automatically provisions infrastructure, deploys workloads, manages TLS certificates, synchronises deployments through GitOps and provides visibility into the health of the environment.</p><p>More importantly, it changed how I think about cloud engineering. I stopped looking at Kubernetes as a collection of tools and started understanding how those tools work together to build reliable platforms.</p><h3>Why I Built This</h3><p>After building my ECS platform using production-grade engineering practices, I wanted to understand what problems Kubernetes was actually designed to solve.</p><p>It’s easy to think Kubernetes is simply another way to run containers.</p><p>The more I learned, the more I realised that containers are only a small part of the story.</p><p>I wanted to understand how networking, ingress, GitOps, certificate management, service discovery and observability all fit together inside a Kubernetes environment.</p><p>Rather than building a complicated application, I chose to build a realistic platform.</p><p>That allowed me to focus on the infrastructure and operational decisions that make Kubernetes environments reliable and maintainable.</p><h3>What the Platform Does</h3><p>The platform provides a secure and automated environment for deploying Kubernetes workloads.</p><p>Traffic is routed to the platform using Amazon Route 53, with ExternalDNS automatically creating and updating the DNS records required to expose the application.</p><p>An AWS Load Balancer forwards requests into the Kubernetes cluster where Traefik manages ingress traffic.</p><p>TLS certificates are automatically issued and renewed by cert-manager.</p><p>DNS records remain synchronized automatically as the platform evolves.</p><p>ArgoCD continuously compares the cluster with the desired state stored in Git and reconciles any differences.</p><p>Prometheus collects metrics while Grafana provides dashboards that make the platform easier to monitor and troubleshoot.</p><p>From the outside, the platform appears straightforward. Behind the scenes, each component has a specific role in keeping the platform secure, automated and reliable.</p><h3>Architecture Principles</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yzmYASoTFBEQii0_wCWH4g.png" /></figure><p>I designed the platform around several engineering principles that I wanted to apply consistently.</p><h3>GitOps</h3><p>Infrastructure changes should be predictable and repeatable.</p><p>ArgoCD continuously synchronises Kubernetes resources from Git, making the repository the single source of truth.</p><p>Rather than manually updating the cluster, deployments happen through version-controlled changes that are easy to review and understand.</p><h3>Secure by Default</h3><p>Security was considered from the beginning rather than added later.</p><p>Worker nodes run within private networking, TLS protects external traffic and GitHub Actions authenticates to AWS using OIDC instead of long-lived access keys.</p><p>IAM permissions follow the principle of least privilege wherever possible.</p><h3>Infrastructure as Code</h3><p>Terraform provisions the AWS infrastructure including networking, IAM, Amazon EKS and supporting resources.</p><p>Helm manages the installation of Kubernetes platform components while Git stores the application manifests.</p><p>Everything can be recreated consistently from code.</p><h3>Operational Visibility</h3><p>Understanding what a platform is doing is just as important as deploying it.</p><p>Prometheus collects metrics, Grafana visualises them and CloudWatch captures logs for troubleshooting.</p><p>This provides multiple ways to understand the health of the environment when something doesn’t behave as expected.</p><h3>Why Amazon EKS</h3><p>Choosing Amazon EKS was a deliberate decision.</p><p>I wasn’t looking for the most complicated technology.</p><p>I wanted to understand how production Kubernetes platforms are typically built and operated.</p><p>Compared with ECS, Kubernetes introduces additional complexity.</p><p>That complexity also provides greater flexibility through its ecosystem of controllers, operators and declarative configuration.</p><p>Building this platform helped me appreciate both approaches.</p><p>Sometimes ECS is the right solution.</p><p>Sometimes Kubernetes is.</p><p>Understanding when to choose each is just as important as knowing how to use them.</p><h3>GitOps with ArgoCD</h3><p>One of my favourite parts of this project was implementing GitOps.</p><p>Instead of treating Git as somewhere to store code, it became the source of truth for the entire platform.</p><p>Whenever changes are committed, ArgoCD detects the difference between Git and the cluster before automatically synchronising them.</p><p>This removes much of the manual work involved in deployments while helping prevent configuration drift over time.</p><p>It also makes Kubernetes deployments visible, reviewable and repeatable.</p><h3>CI/CD</h3><p>Automation became an important part of the platform.</p><p>The GitHub Actions pipelines build container images, perform security scanning with Trivy, push approved images to Amazon ECR, validate infrastructure changes with Terraform and deploy updates through automated workflows.</p><p>Authentication to AWS uses OIDC federation, eliminating the need for long-lived credentials inside GitHub.</p><p>The goal wasn’t simply to automate deployments.</p><p>It was to create confidence that every deployment followed the same secure and repeatable process.</p><h3>Kubernetes Add-ons</h3><p>Kubernetes provides the foundation, but the surrounding ecosystem is what makes a platform operational.</p><p>Traefik manages ingress traffic.</p><p>cert-manager automates certificate management.</p><p>ExternalDNS keeps DNS records synchronised.</p><p>Prometheus collects metrics.</p><p>Grafana provides operational dashboards.</p><p>Each component solves a specific problem, and together they create a platform that is easier to operate and maintain.</p><h3>What I Learned</h3><p>The biggest lesson from this project had very little to do with Kubernetes commands.</p><p>I learned that Kubernetes is an ecosystem rather than a single technology.</p><p>Networking became even more important than I expected.</p><p>GitOps simplified deployments by making Git the source of truth.</p><p>Observability proved to be something that should exist from the very beginning instead of being added later.</p><p>Most importantly, I realised that successful platforms are built from many small engineering decisions that work together rather than one impressive piece of technology.</p><h3>What I’d Improve Next</h3><p>If I continued evolving this platform, the next steps would include:</p><ul><li>Implementing Cluster Autoscaler.</li><li>Introducing Horizontal Pod Autoscaling.</li><li>Expanding monitoring and alerting.</li><li>Strengthening disaster recovery capabilities.</li><li>Exploring progressive delivery strategies such as blue-green or canary deployments.</li><li>Continuing to improve platform resilience and automation.</li></ul><p>The platform was designed so these improvements could be added without significant architectural changes.</p><h3>Final Thoughts</h3><p>This project wasn’t about learning Kubernetes commands.</p><p>It was about understanding how networking, automation, security, observability and GitOps come together to build a reliable platform.</p><p>Building this platform gave me a much better appreciation of the engineering decisions that happen behind the scenes of modern cloud-native environments.</p><p>The biggest takeaway wasn’t learning another technology. It was understanding how many small engineering decisions come together to build a platform that is secure, maintainable and designed to evolve over time.</p><p><strong>GitHub Repository:</strong> <a href="https://github.com/Param2ray/eks-production-platform"><em>Project Link</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e27d533115ce" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building a Production-GradeHealth Monitoring Platform on AWS with ECS Fargate]]></title>
            <link>https://medium.com/@param_2ray/building-a-production-style-health-monitoring-platform-on-aws-with-ecs-fargate-3fd39784f5fc?source=rss-d9df9d97174a------2</link>
            <guid isPermaLink="false">https://medium.com/p/3fd39784f5fc</guid>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[terraform]]></category>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Paramjyot Singh Tooray]]></dc:creator>
            <pubDate>Fri, 17 Apr 2026 01:09:00 GMT</pubDate>
            <atom:updated>2026-10-27T22:02:51.586Z</atom:updated>
            <content:encoded><![CDATA[<h3>TL;DR</h3><p>I built a production-grade health monitoring platform on AWS using ECS Fargate, Terraform, GitHub Actions, Amazon Route 53, and Gatus.</p><p>The goal was not just to deploy an application. It was to design and operate an end-to-end cloud platform using real engineering practices around security, automation, observability, and reliability.</p><p>The result is a secure HTTPS endpoint, private container workloads, automated CI/CD pipelines, infrastructure as code, and real-time health monitoring.</p><p>This project helped me think less like someone following tutorials and more like someone responsible for running systems properly.</p><h3>Why I Built This</h3><p>A lot of cloud projects stop at “it works.”</p><p>The application runs. The page loads. The task completes.</p><p>But production systems require more than that.</p><p>They need to be secure. Repeatable. Observable. Easy to change. Safe to operate.</p><p>I wanted to build something that reflected those expectations.</p><p>Instead of focusing on a complex application, I focused on the platform around it.</p><p>That meant thinking about networking, deployment flow, IAM design, logging, HTTPS, container hardening, and operational clarity.</p><h3>What the Platform Does</h3><p>The application uses Gatus, a lightweight monitoring tool that checks the health of endpoints and services.</p><p>Users access the platform through a custom domain over HTTPS.</p><p>Amazon Route 53 resolves the custom domain and routes traffic to an AWS Application Load Balancer over HTTPS.</p><p>Requests are then forwarded to ECS Fargate tasks running in private subnets.</p><p>Those containers run health checks against internal and external services and present the results through a dashboard.</p><p>Logs are sent to CloudWatch for visibility and troubleshooting.</p><p>From the outside, it looks simple.</p><p>Behind the scenes, every layer was designed intentionally.</p><h3>Architecture Principles</h3><p>I built the platform around a few core principles.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*67lewo5HW0WU9l1ZShsYwQ.png" /></figure><h3>1. Private Workloads</h3><p>The containers run in private subnets with no direct inbound internet access.</p><p>Only the load balancer is public.</p><p>This reduces exposure and follows common production patterns.</p><h3>2. Secure by Default</h3><p>HTTPS is enabled using AWS Certificate Manager.</p><p>Amazon Route 53 manages DNS.</p><p>GitHub Actions uses OIDC federation instead of long-lived AWS credentials.</p><p>IAM roles are scoped to the minimum permissions required.</p><h3>3. Infrastructure as Code</h3><p>The full environment is defined in Terraform.</p><p>Networking, compute, IAM, ACM certificates, Route 53 DNS, and deployment resources are all managed through Terraform.</p><p>That means repeatability, version control, and cleaner change management.</p><h3>4. Operational Visibility</h3><p>CloudWatch receives container logs automatically.</p><p>The load balancer performs health checks.</p><p>Gatus provides live service status.</p><p>If something breaks, there is a clear place to investigate.</p><h3>Why ECS Fargate</h3><p>Choosing ECS Fargate was deliberate.</p><p>I wanted to focus on delivering and operating workloads rather than managing servers.</p><p>Fargate removes the need to patch instances, manage worker nodes, or maintain cluster capacity.</p><p>It integrates cleanly with IAM, ALB, CloudWatch, and ECR.</p><p>For this use case, it was the right level of abstraction.</p><p>The goal was not complexity.</p><p>The goal was sound engineering decisions.</p><h3>Terraform Design Matters</h3><p>Terraform is not just about writing resources.</p><p>Structure matters.</p><p>I split the infrastructure into modules with clear responsibilities.</p><p>One module handles networking.</p><p>One handles the load balancer.</p><p>One handles ECS.</p><p>Others manage IAM, certificates, and DNS.</p><p>This makes the code easier to understand, safer to change, and easier to scale over time.</p><p>I also used remote state in S3 with locking enabled to support safe automation workflows.</p><p>I also separated one-time foundational resources such as the remote state backend, ECR, and IAM into a bootstrap layer.</p><h3>CI/CD as a Control System</h3><p>The CI/CD pipelines were designed to do more than deploy code.</p><p>They act as quality and security gates.</p><p>The workflows:</p><ul><li>Build the Docker image</li><li>Tag images using commit SHA</li><li>Scan images with Trivy</li><li>Scan Terraform with Checkov</li><li>Push approved images to ECR</li><li>Run Terraform plan</li><li>Apply infrastructure changes through a manually approved workflow</li><li>Run post-deployment health checks</li></ul><p>Container images are scanned with Trivy before deployment, while Checkov surfaces infrastructure security findings during the pipeline for review.</p><p>This provides automated quality and security feedback while keeping the deployment workflow visible and auditable.</p><h3>Container Optimization</h3><p>I used a multi-stage Docker build and a minimal runtime image.</p><p>The container runs as a non-root user with a smaller attack surface.</p><p>One of the most satisfying improvements was reducing image size from 2.55 GB to 80.4 MB.</p><p>That is roughly a 97 percent reduction.</p><p>Smaller images mean faster pulls, quicker deployments, and fewer unnecessary components in production.</p><h3>What I Learned</h3><p>This project reinforced several lessons.</p><h3>Simpler Is Often Better</h3><p>Good engineering is not adding more services. It is choosing what is necessary and operating it well.</p><h3>Security Must Be Built In</h3><p>IAM, secrets handling, HTTPS, and pipeline authentication should not be afterthoughts.</p><h3>Observability Matters Early</h3><p>Logging and health checks are not extras. They are part of running systems responsibly.</p><h3>Real Confidence Comes from Ownership</h3><p>Building everything end to end forces deeper understanding. You stop memorising tools and start understanding systems.</p><h3>What I’d Improve Next</h3><p>If I continued evolving this platform, the next steps would be:</p><ul><li>Multi-environment deployments</li><li>Auto scaling policies</li><li>Metrics and alerting</li><li>Blue green or canary deployments</li><li>More advanced monitoring integrations</li><li>Promotion-based release flows</li></ul><p>The foundation was built so those improvements can be added cleanly.</p><h3>Final Thoughts</h3><p>This project was never about creating the most complicated architecture.</p><p>It was about building a secure, automated and maintainable platform by applying production-grade engineering practices.</p><p>That is what good platforms aim to be.</p><p>And for me, that is where cloud engineering becomes genuinely interesting.</p><p><strong>GitHub repository: </strong><a href="https://github.com/Param2ray/ecs-production-healthcheck-service">Project Link</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3fd39784f5fc" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>