◆ প্রজেক্টLive
2019 to today · Learning & Practicing
Production Cloud Platform
Six years owning a production AWS environment end-to-end. Architecture, deploys, cost, scaling, security. The platform behind everything else on this page.
AWSEKSRDSECRCloudflareArgo CDTerraform
Context
The platform isn't just the application. It's the cloud underneath. I've owned a production AWS environment end-to-end for 6+ years: architecture, deploys, cost, scaling, security.
Scope
- Application layer: Laravel + NestJS + Next.js (see E-commerce Platform Rebuild)
- Microservices on EKS: the Go services (Merchant Center Feeds, Provider Price Sync, Product Feeds Generator) run here
- Data: RDS Multi-AZ writer + read replicas, RDS Proxy for connection pooling, ClickHouse for log/analytics offload
- Edge: Cloudflare in front, WAF tuned by hand, custom rules for bot defense
- CI/CD: GitHub Actions, ECR, Argo CD, Terraform for infra
Cost-first mindset
The default reaction to load is to upsize. I prefer attacking the workload. A few examples:
- Logs to S3 + ClickHouse instead of bigger RDS storage
- Redis cache for hot paths to keep the DB out of the request loop where possible
- N+1 fixes at code level as the first lever, hardware as the last
Security
- Defended the platform against multi-million-hit bot / DDoS attacks. Cloudflare WAF set up and tuned myself.
- Run my own pentests against the platform. Real findings: cart session hijacking, missing rate limits on forgot-password and DB-heavy endpoints. All fixed.
- CEH v12 course completed (EC-Council). Applied mindset, not just paper.
Stack
AWS (EKS, RDS, S3, SQS, ECR), Cloudflare, Argo CD, Terraform, GitHub Actions.
Outcome
Same scale, same reliability, smaller AWS bill. The default reaction to load is to upsize. I prefer attacking the workload first.
flowchart TB
title["⚠ Default reaction"]
L1[High DB load] --> U1[Upsize RDS]
U1 --> $[Bigger bill]
$ -.->|"recurring"| MORE[More upsizing]
style title fill:transparent,stroke:transparent
flowchart TB
title["✓ Cost-first"]
L2[High DB load] --> A1[Logs to S3 + ClickHouse]
L2 --> A2[Redis cache hot paths]
L2 --> A3[Fix N+1 in code]
A1 --> SAVE[Same RDS, lower bill]
A2 --> SAVE
A3 --> SAVE
style title fill:transparent,stroke:transparent