TONY VERIN — DEVOPS

I build the pipeline, guard the deploy, and keep a cool head when it breaks.

PROFILE

Resilient systems, built from scratch.

I'm a DevOps and Cloud Infrastructure Engineer who loves building resilient, automated systems from scratch — containerization through full CI/CD pipelines. Self-driven learner, sharp troubleshooting instincts, and a passion for owning the entire deployment process. When things break under pressure I keep a cool head — a skill mastered as an IDF sniper and spotter, where absolute focus and precision were everything.

  • ▪AWS Solutions Architect Associate (SAA-C03) — IITC College
  • ▪DevOps Engineer Certificate — IITC College
  • ▪Magshimim — Advanced Software Engineering & Networking
CLOUD & INFRASTRUCTURE
AWS (EC2, ALB, CloudFront, EKS, ECS, Bedrock, IAM…) · Terraform
CONTAINERS & CI/CD
Docker · Kubernetes (EKS, k3s, Minikube) · Helm · ArgoCD (GitOps) · GitHub Actions (OIDC auth, reusable workflows)
OBSERVABILITY & SCRIPTING
Prometheus · Grafana · Python (FastAPI) · Linux
LANGUAGES
Hebrew · English · Russian

WORK

Two systems, owned end to end.

FLAGSHIP

DeployGuard

git init → safely-deployed k8s in 4 commands.

  • Python CLI (`dg`, on PyPI as dg-deploy, also on Homebrew) that turns Kubernetes deployments into a guided four-command workflow — no manual, unvalidated configs.
  • Two-stage safety engine: blue/green health pre-check before traffic shifts, then a gradual canary rollout (10% / 50% / 100%) with automated rollback on Prometheus error spikes.
  • Policy-as-code guard layer — kubeconform + Trivy + 7 custom safety rules, packaged as both a CLI stage and a shareable Claude Code skill.
  • Config merges CLI flag → project file → personal file → built-in defaults, so a team commits one behavior and a solo dev needs zero config.
  • Runs on k3s over AWS EC2 Spot instances to keep the whole environment around $6/mo, with `make pause` dropping it to ~$3.55/mo without losing the DNS/EIP.
  • Python
  • Typer
  • Rich
  • Kubernetes
  • k3s
  • Terraform
  • Prometheus
  • Trivy
  • AWS
Case study →GitHubPyPI

PROJECT

4RCH3R · Investment Terminal

An AI strategist reasoning over your live holdings.

  • Solo-built full-stack, multi-user portfolio tracker — FastAPI + MongoDB Atlas with JWT auth.
  • AI assistant on AWS Bedrock (Claude) with Guardrails, reasoning over live holdings plus real-time web search.
  • One chat endpoint normalized across 4 LLM backends (Bedrock, Groq, LM Studio, Ollama) into a single response shape.
  • Terraform-managed AWS infrastructure with a tag-gated GitHub Actions deploy pipeline; live prices via yfinance, sector analysis, NAV heartrate chart.
  • FastAPI
  • MongoDB Atlas
  • AWS Bedrock
  • Terraform
  • GitHub Actions
  • yfinance
  • Groq
  • Ollama
Case study →GitHub

TELEMETRY

Measured, not promised.

commands, git init → live k8s
4
policy guard rules in the deploy path
7
canary steps — 10 / 50 / 100
3
a month for the full k3s environment
$6
LLM backends, one response shape
4

CONTACT

Need someone who owns the whole deploy?

toni7891@gmail.com