portfolio.sh
$ _
 ____             ___              
|  _ \  _____   _/ _ \ _ __  ___   
| | | |/ _ \ \ / / | | | '_ \/ __|  
| |_| |  __/\ V /| |_| | |_) \__ \  
|____/ \___| \_/  \___/| .__/|___/  
                       |_|          
          

Infrastructure as Code | Kubernetes | Cloud Native

scroll

# Tech Stack

[1] Container and Orchestration

Kubernetes Docker Helm ArgoCD Harbor

[2] Cloud and IaC

AWS Terraform Terragrunt GitHub Actions

[3] Networking

Calico CNI MetalLB Site-to-Site VPN gRPC

[4] Database and Storage

MariaDB MaxScale PostgreSQL Redis Cluster Ceph/Rook

[5] Observability

Prometheus Grafana Loki AlertManager

[6] AI/ML Infra

NVIDIA Triton GPU Optimization Model Serving

# Projects

01

Air-gapped AI Platform

Offline AI inference platform with triple-redundant clustering and high availability architecture

INGRESS
MetalLB
NGINX
SERVICE
Triton Server
Triton Server
Backend API
DATA
MariaDB
MaxScale 3-node
Redis
6-node Cluster
SeaweedFS
3x Replication
99.99% Availability
5,000 req/sec
85% GPU Utilization
<30s MTTR
02

IaC Pipeline with Terragrunt

DRY-principle based multi-environment infrastructure automation with GitOps pipeline

terragrunt.hcl
# Stage/Prod 95% code reuse
include "root" {
  path = find_in_parent_folders()
}

terraform {
  source = "..//modules/ecs-cluster"
}

inputs = {
  environment    = "production"
  instance_count = 3
  enable_ha      = true
}
PR Created
TF Plan
OK Review
GO Apply
80% Code Reduction
96% Deploy Time Down
Zero Env Drift
03

Hybrid Cloud Architecture

On-premise and AWS hybrid setup with dual VPN and cost optimization

AWS VPC
Route53 ALB EC2
IPSec VPN x2
On-Premise K8s
MetalLB Ingress GPU Pods
74% Cost Reduction
$383 Monthly Savings
04

Container Registry Migration

Zero-downtime migration from Docker Registry to Harbor with security scanning

Source
Docker Registry
2,847 images
parallel sync
Target
Harbor + Trivy
vuln scan
85% Time Reduction
90% Automation
$2,880 Annual Savings
05

Observability Stack

Loki + Promtail + Grafana + AlertManager based log monitoring system

LOG JSON Logs
IN Promtail
DB Loki
VIZ Grafana
! Slack
92% MTTR Reduction
<60s Detection
99.8% Availability
06

EKS GitOps Automation

Terraform + ArgoCD based EKS cluster and AI workload management

GitHub
GitHub Actions
infra/
Terraform
EKS Cluster
k8s/
ArgoCD
Workloads
On-Demand (stability)
Spot Instance (60% off)
GPU Node (NVIDIA)