Case Study · Cloud Migration
Migrating a SaaS Platform from Docker Compose to ECS Fargate
Re-architected a two-instance Docker Compose monolith into an 8-microservice architecture on ECS Fargate: Secrets Manager for scoped config, TLS-enforced ElastiCache Redis, an NLB path for RTMP streaming traffic, and SES for transactional email.
Impact: Went from 2 monolithic EC2 instances to 8 independently deployable services, with secrets no longer stored in plaintext anywhere in the stack.
ECS FargateElastiCache RedisSecrets ManagerSESNLB
Read the full case study →
Case Study · Infrastructure as Code
Cutting Infrastructure Costs with a Multi-Cloud Comparison
Provisioned a full 10-resource production architecture (VPC, managed Postgres/Redis, autoscaling API and worker tiers, CDN, load balancer), then benchmarked equivalent specs across an Indian cloud provider, AWS, and GCP to give the client real INR pricing instead of USD list prices.
Impact: Cost report found the Indian cloud provider 50-70% cheaper than AWS/GCP at equivalent specs, giving the client a concrete number to base the infrastructure decision on.
E2E NetworksTerraformAutoscalingCloudflare
Read the full case study →
Enterprise Gaming Platform Migration
Migrated a real-time multiplayer gaming platform (20+ containerized microservices) from EC2 to Amazon ECS Fargate with zero downtime. Blue-green and rolling-update deployments, automated CI/CD with Drone CI, standardized infra with Terraform.
Impact: Delivered 85% faster releases and 40% infrastructure cost savings with zero downtime during migration.
ECSEC2TerraformBlue-Green DeploymentsDrone CI
Production MLOps Pipeline on AWS
Built an end-to-end MLOps platform: S3 data lake versioned with DVC, Kubeflow training pipelines with MLflow tracking, FastAPI inference via KServe, Evidently AI drift monitoring, and GitHub Actions CI/CD with automated model quality gates.
Impact: Enabled experiment-to-production in under 2 hours, with drift alerts firing within 5 minutes of distribution shift.
MLflowDVCKubeflowKServeEvidently AI
Production RAG Chatbot on Secured Kubernetes
Built a production RAG chatbot: PDF ingestion, Qdrant vector database, Ollama LLM serving, FastAPI streaming API on k3s with MetalLB. Security layer: OPA/Gatekeeper, Cosign image signing, Gitleaks and Trivy scanning.
Impact: p99 query latency under 800ms with zero secrets committed to the repo.
LangChainQdrantOllamak3sOPA/GatekeeperCosign
Secure DevSecOps CI/CD Pipeline
Production-grade DevSecOps pipeline: SAST (Semgrep), SCA (Snyk), Trivy container scanning, SBOM generation (Syft), Cosign signing, Gitleaks, OWASP ZAP DAST, and OPA/Gatekeeper policy enforcement. Blocks merge on any HIGH/CRITICAL finding.
Impact: Zero HIGH/CRITICAL findings reached production, with a full audit trail for every build.
SemgrepSnykTrivySyftCosignOWASP ZAP
Multi-Environment Cloud Infrastructure Platform
Designed and maintained a reusable Terraform module library covering 20+ AWS and GCP resource types (VPC, ECS, RDS, S3, IAM, GKE, Cloud Run), standardizing dev, staging, and production across 15+ projects.
Impact: Provisioning time cut ~80%, from days to under 2 hours, with zero configuration drift.
TerraformAWSGCPTerraform CloudAnsible
Azure to GCP Migration (Ruby on Rails)
Migrated a multi-tier Ruby on Rails application from Azure VMs to Google Cloud Run: containerized deployments, private VPC networking, Cloud SQL, Workload Identity Federation for keyless auth, and IAP for zero-trust access.
Impact: Zero-downtime migration with 30% lower monthly cloud spend and static keys eliminated entirely.
Cloud RunCloud SQLIAPWorkload Identity Federation
