I build production-grade cloud infrastructure and agentic AI pipelines on AWS.
My background is unusual for a DevOps engineer: I have a B.Sc in Biochemistry and an M.Sc in Biotechnology. That scientific background shaped how I approach infrastructure -- methodical, documented, precise. If it cannot be reproduced exactly, it is not done right.
My most recent project is a complete agentic DevOps pipeline: Claude Code Skills for Terraform scaffolding, planning, applying, and deploying; three SubAgents for security auditing, infrastructure writing, and cost optimisation; MCP server integration for live provider schemas; and a production safety layer of hook scripts that block dangerous commands before they execute.
"Safety is architecture, not discipline."
A production-grade agentic DevOps pipeline built from scratch -- environment to safety hooks.
Environment Setup → CLAUDE.md → Skills → Live Deploy → SubAgents → MCP → Safety Hooks
| Component | What It Does |
|---|---|
| CLAUDE.md | Persistent project memory -- agent knows architecture, conventions, constraints |
| /scaffold-terraform | Generates complete Terraform config from template spec |
| /tf-plan | Validates, plans, scans for destructions, returns plain-English summary |
| /tf-apply | Applies saved plan -- provisions real AWS resources |
| /deploy | Syncs to S3, triggers CloudFront invalidation, reports live URL |
| security-auditor | Read-only SubAgent -- audits Terraform for misconfigurations |
| tf-writer | Read-write SubAgent -- generates Terraform using live MCP schema |
| cost-optimizer | Read-only SubAgent -- reviews infrastructure for cost inefficiencies |
| SAY hook | Blocks destructive prompts before Claude processes them |
| DO hook | Blocks dangerous commands before they execute |
| LOG hook | Writes timestamped entry to deploy.log on every terraform apply |
Live site: viviancloud.site | Tech: Claude Code • Terraform • AWS S3 • CloudFront • MCP • Bash • WSL2
A real-world infrastructure setup -- 2 web servers, 1 Nginx load balancer, Elastic IP, DNS routing, health checks.
GitHub webhook, automated build, SSH transfer to NFS, full documentation.
Custom VPC across 2 AZs, Next.js on EC2 behind ALB, Node.js backend on private EC2, Amazon RDS with Read Replica, AWS SSM Session Manager for zero open-port access.
I document everything publicly -- real projects, real problems, real fixes.
- Portfolio: viviancloud.site
- dev.to: dev.to/vivian_okose
- Hashnode: hashnode.com/@vivianokose
- Medium: medium.com/@vivianokose
- LinkedIn: linkedin.com/in/okosechiamaka
"Safety is architecture, not discipline."
Available for DevOps engineering roles, cloud consulting, and AI automation projects. Remote-first.
