DevOps / Platform Engineer · Chicago

Vera Terna

I run AWS + Kubernetes platforms — and build the AI agents on top.

Five years operating EKS at scale across 7 AWS accounts, delivered with Terraform and GitOps — and agentic AI built into the platform itself.

CKA CKAD AWS · EKS · Terraform · Argo CD · Datadog
this site's real pipeline — click a node
Repo source of truth CI · Actions OIDC keyless Terraform IaC S3 origin CloudFront edge · TLS Route 53 veraterna.com Visitor 🌐 you
Tip: hover or tap any node — this is the real deploy path that ships and serves the page you're reading.

By the numbers

Platform work, quantified

7
AWS accounts, multi-region
dozens
EKS clusters in the fleet
18→6
min alert→cause (AI)
30
repos w/ AI PR-review
<24h
infra drift caught
3
environments

Before / after

What the platform work moved

CI runner cold pull

before55s
after10s

SOCI + 600 MiB/s gp3, validated over 3,200+ cold pulls

Prod compute spend / month

beforebaseline
after−25%

Spot + Graviton + Karpenter consolidation (node util 45%→65%)

Cluster bootstrap

before~2h
after~20m

Reusable Terraform modules + Argo CD reconciliation

The stack

What I build with

☁️

Cloud & AWS

EC2, Spot, Graviton, CloudFront, ALB, Route 53, WAF — multi-account, multi-region.

⎈

Kubernetes & GitOps

Amazon EKS, Karpenter, Argo CD / ApplicationSets, Helm.

📐

IaC & CI/CD

Terraform, GitHub Actions, drift detection, OIDC keyless deploys.

📊

Observability

Datadog, OpenTelemetry, CloudWatch, incident response & triage.

🤖

AI & Agentic

Claude Code, MCP, custom skills, context engineering (CLAUDE.md / AGENTS.md).

🔒

Foundations

Docker, Linux, Bash, least-privilege IAM, secret scanning.

The centerpiece

AI agents I built — shown running

★ FLAGSHIP

AI PR-review agent

Claude Code as a customized GitHub Action — reviews every diff before a human does.

PR opened→ diff + Jira req→ secret scan→ inline review
30repositories under continuous review

● redacted real output — shown, not described

Incident-diagnostics workflow

A read-only agent that takes on-call from alert to probable cause with evidence attached.

alert→ Datadog + K8s + Argo→ deploy history→ probable cause
18→6min, alert to probable cause

● read-only & safe — remediation stays with the engineer

Track record

Experience

Platform Engineer · AppsFlyer

Sep 2024 — Present

Global mobile-attribution & marketing-analytics SaaS · Remote

  • Own the EKS platform behind a fleet of dozens of clusters across 7 AWS accounts and regions, on reusable Terraform modules and Argo CD ApplicationSets.
  • Cut CI cold pulls ~55s → ~10s (SOCI + tuned gp3) across 3,200+ production pulls.
  • Cut production compute spend ~25% with Spot, Graviton, and right-sized Karpenter node pools.
  • Built an AI PR-review agent across 30 repos and a read-only incident-diagnostics workflow.

DevOps Engineer · Kensium Solutions

Sep 2021 — Jul 2024

E-commerce & ERP systems-integration consultancy · Chicago, IL

  • Reusable Terraform modules stood up a full client environment in under 2 hours vs 1–2 days — across 8 clients.
  • Rebuilt the release pipeline in GitHub Actions — ~35 → ~12 min, kept a prod approval gate.
  • Standardized edge routing & security (CloudFront, ALB, Route 53, WAF) across clients.

Get in touch

Let's talk platforms & agents