Cloud Engineer

CLOUDMILE

CLOUD INFRASTUCTURE Dept.

Mar 2022 -  Dec 2022
Taipei, Taiwan

• Terraform Infra CICD Pipeline (Manufacturer)

Gitlab / GCS / GKE gitlab runner

1. push code to Gitlab repo
2. merge request to trigger CI (main)
3. CI : (different folder each main branch)
   • terraform scan / init / validate / plan
4. merge main branch reviewer manual trigger CI 
5. CI : (different folder each main branch)
   • terraform scan / init / validate / plan / apply
6. gitlab runner executor pod start worker pod build common infra

• Official Website GCP SOLUTION (Manufacturer)

VPC - Custom mode / Subnet / Firewall Rule (network)

• VPC network

Wordpress / LB / WAF Armor/ DNS / Unmanaged Instance Group / GCE / CloudSql / GCS / CDN (web service)

• Build web server
• Able to access the web server (Wordpress) through custom domains
• Implement security WAF solution - Armor
• Build wordpress db
• Mount additional storage - GCS
• Cache static content

• MLaaS GCP SOLUTION (Bank)

HA VPN / Shared VPC / Subnet / Cloud Router / GKE IP Range / Cloud SQL IP Range / Redis IP Range (hybrid, centralized network)

• SharedVPC network
  
• Hybrid Cloud
• Subnet IP Range Planning
  • K8s Control Plane  
  • K8s Worker Nodes Primary IP range
  • K8s Worker Pod, Service Secondary  IP range
  • Cloud SQL, Redis (psc)

LB / DDOS Armor / DNS / Unmanaged Instance Group / logging / GCS / GKE / CloudSql / DMS (ml api )

• HTTPS LB 
• Implement security solution - Armor
• Able to access the MLaaS through custom domains
• URL map redirect to ingress Nginx reverse proxy, then route to different service
  
• Log management retention requirements (import to GCS)
• Start the "model training" job on-prem, and use the GPU to train the model on the GKE
• Build ML training model db
• DB continuously migrate

AzureDevOps / Artifact Registry / Argo CD (application cicd)

1. push code to Azure DevOps repo
2. merge request to trigger CI
3. CI :
   1. unit test
   2. build ML model training into the container image
   3. push the container image to the artifact registry
4. CD :
   1. apply application on argoCD
   2. argoCD will sync of k8s manifests on the Azure DevOps repo with GKE via tracking and deployment strategies
   3. start ML model training task on GKE

• BAS ETL GCP SOLUTION (Telecom Co.)

HA VPN / Shared VPC / Subnet / Cloud Router / Firewall Rule (hybrid, centralized network infra)

• Shared VPC network
  
• Hybrid Cloud

Dataflow / GCE / GCS / SNAT / Custom Route / Informix / BQ (jdbc data flow)

• Dataflow uses JDBC to connect to the Informix
• Dataflow template to copy data from Informix with available JDBC drivers into BQ
• Worker node with JDBC driver installed to rewrite original IP address through SNAT 
  

• AWS Migrate to GCP  Project 

      (Food Delivery Platform)

VPC / Subnet / NAT / VPC peering / Firewall Rule (network)

• STG / PRD env VPC network 

LB / Unmanaged Instance Group / Nginx /  GCE  (network proxy)

• Deploy web server 
• To connect to a web server through Nginx

Jenkins / GCE (cicd)

• Migration of Jenkins CI/CD
• Deploying ReactJS and Java application to App Engine using Jenkins.

App Engine / MongoDB / GCE / DMS / CloudSql / Redis / private service access (web application)

• Deploy Application 
• Migrate Beanstalk App to GCP App Engine 
• ReactJS and Java Application connect to DB and Redis