Senior DevOps · SRE · Platform Engineer. Founder of Orrala Systems.

I have spent 18+ years making infrastructure boring on purpose: secure, reliable, reproducible, and cheap to run. I like owning the whole lifecycle, from the architecture diagram to the 3am page to the end-of-month cost report. Lately I build the cloud-native environments and automated quality frameworks that stress-test advanced AI systems, plus the agentic tooling that handles the tedious parts for me.

• Architect and run production platforms on AWS and Kubernetes, codified end to end in Terraform.
• Ship with CI/CD and GitOps that move fast without breaking prod.
• Harden the boring-but-critical layer: IAM, secrets, RBAC, Pod-Security, supply-chain, observability, SLOs.
• Chase cost and reliability: right-size, decommission, and automate the toil out of existence.

• Cut cloud spend ~25% (six-figure annual savings) by right-sizing and retiring legacy AWS workloads.
• Rebuilt CI/CD from scratch and dropped time-to-production by 60%.
• Ran a zero-downtime GitHub Enterprise to GitHub.com migration across distributed teams.
• Built the platform behind a nationwide pharmacovigilance system handling ~1M reports a year.
• Owned the technical quality bar for 40+ production-realistic infrastructure environments and authored ~100 evidence-based technical reviews.

Two end-to-end builds you can actually read:

• Example Voting App on AWS EKS: Terraform (EKS, IRSA, ECR, EBS CSI), a custom Helm umbrella chart, and GitHub Actions OIDC CI/CD with no static secrets.
• FastAPI productionized on Kubernetes: Helm umbrella chart, async PostgreSQL, and Prometheus plus Grafana, self-contained in a single k3d container.

Cloud AWS (EKS, EC2, IAM, VPC, Fargate, SSM, CloudWatch), Azure Orchestration Kubernetes, Docker, Helm IaC Terraform, Terraform Cloud Delivery GitHub Actions, Jenkins, Travis CI, Harness.io, Argo CD Reliability & Security Prometheus, Grafana, distributed tracing, RBAC, Pod-Security, supply-chain security Languages Python, Bash, TypeScript

• CNCF Kubestronaut track: KCNA, KCSA, CKA, CKAD, CKS
• AWS SysOps Administrator (Associate) → DevOps Engineer (Professional)
• Expert Certificate in Cybersecurity, MIU City University Miami

• M.S. in Software Development & Operations (DevOps), dual credential (México + Spain, UNIR), 2026
• AWS Certified Cloud Practitioner · HashiCorp Terraform Associate (003) · TOEIC
• Mensa México member

A homelab I run like production, and a few AI and automation builds that got out of hand. I build them spec-first, with AI agents in the loop:

• Agentic automation, treated like prod. LLM-agent pipelines that turn messy real-world input into a finished result with no babysitting: OCR, agents that drive third-party web portals, and a cost-tiered model cascade that only escalates to the expensive model when the cheap path fails. The one I depend on daily runs on a typed, Postgres-backed orchestrator I wrote to replace a no-code tool, with SOPS-encrypted secrets, restic backups, Ansible, and CI with secret scanning. Anything I rely on should be boring and observable.
• A council of AI models that argues back. Several models take debating personas, reason across rounds, vote, and converge on an answer instead of just agreeing with me. Built to fight model sycophancy, exposed as both an MCP server and a CLI.
• First place, Capital One México track, Genius Arena 2026. With team Eonix I shipped a cloud-native, AI-powered financial-education platform, MVP built end to end with full spec-driven development on AWS serverless.
• The homelab itself. A GitOps control plane, documented runbooks, Intel-accelerated camera transcoding, and observability, read-only by default. If a change is not reproducible, it does not ship.
• The terminal. Six-plus machines across macOS, Linux, and Windows that look identical because I wrote a small cross-OS theming engine to keep them that way, down to a low-blue-light palette I tuned by hand for long nights.

If it can be automated, I have probably already automated it.

• LinkedIn: Christian Orrala
• Building, scaling, or rescuing infrastructure? Orrala Systems takes on select engagements.
• Or just open an issue or discussion here.

Less "it works on my laptop," more "production-ready and disposable."

_{🕉 Om, the sound of the universe.}