# Gergely Peidl > Lead DevOps and Cloud Infrastructure Engineer in Budapest, Hungary. 13+ years in software and infrastructure engineering since October 2013, including 10+ years in DevOps and Cloud Infrastructure since March 2016. Started as a software engineer with frontend and backend responsibilities before moving toward infrastructure. Focus: AWS, Kubernetes, infrastructure as code, CI/CD, automation. Runs a Kubernetes homelab that is an ongoing project, and a smart home running Home Assistant and ESPHome. ## Site - [Homepage](https://peidl.net/): the whole profile, one static HTML page - [About](https://peidl.net/#about): who I am and what I work on - [Experience](https://peidl.net/#experience): roles, dates, and outcomes - [Stack](https://peidl.net/#stack): the tools I actually run in production - [Homelab](https://peidl.net/#homelab): the Kubernetes cluster and its case studies - [Automation](https://peidl.net/#automation): Home Assistant, ESPHome, and a fleet of 50+ devices - [Contact](https://peidl.net/#contact): email, GitHub, and LinkedIn ## Case studies Each card on the homepage links to a decision record in the planet-express repo. Status is stated on the card. The three delivered ones are also published as FAQPage structured data on the homepage, so an answer engine can quote them and cite the ADR. - [Why Talos, not k3s (ADR-001)](https://github.com/geripgeri/planet-express/blob/main/docs/decisions/ADR-001-kubernetes.md): delivered. Minimal, immutable, API-managed nodes, provisioned by OpenTofu. No SSH, no shell, no package manager, so the whole node is defined in git. k3s makes setup faster by bundling components and relaxing defaults, which suits getting started but not a system meant to mirror production. - [One public repo, no home network (ADR-021)](https://github.com/geripgeri/planet-express/blob/main/docs/decisions/ADR-021-public-mirror-privacy-partitioning.md): delivered. A mirror workflow reads export-ignore paths from .gitattributes and strips them from the published history, then gitleaks scans what is left. It aborts rather than publish if the exclusion list comes back empty. - [Secrets you can still read in review (ADR-009)](https://github.com/geripgeri/planet-express/blob/main/docs/decisions/ADR-009-sops.md): delivered. SOPS with age encrypts values and leaves YAML field names in plaintext, so a secret change reads as a one-line diff instead of an opaque blob. Kubernetes Secrets are decrypted inside the ArgoCD repo-server by ksops. - [Paging through Signal (ADR-012)](https://github.com/geripgeri/planet-express/blob/main/docs/decisions/ADR-012-alerting.md): in progress, the decision record exists but implementation is not complete. The planned design routes Alertmanager to Signal, which pushes to my phone. Email waits until I open my inbox, which is wrong for a page. - [All decision records](https://github.com/geripgeri/planet-express/tree/main/docs/decisions): the full set ## Projects - [planet-express](https://github.com/geripgeri/planet-express): the homelab. Talos Linux and ArgoCD running on a single bare-metal host, an Architecture Decision Record for every non-obvious choice, the whole Kubernetes cluster under infrastructure as code ## Facts - Current role: Lead DevOps and Cloud Infrastructure Engineer (level 4 of 5, the Staff-equivalent band) - Current company: Protex AI - Recognition: Star Employee Award, Protex AI Outstanding Contributor Awards, September 2026 - Location: Budapest, Hungary (CET/CEST) - Total engineering experience: 13+ years, since October 2013 - DevOps and Cloud Infrastructure experience: 10+ years, since March 2016 - Early career: software engineer with frontend and backend responsibilities - Core stack: AWS, Kubernetes (Talos Linux), OpenTofu / Terraform, Terragrunt, Ansible, ArgoCD, Prometheus, Grafana, Loki, Tempo, Cilium, Kyverno - CI/CD background: Jenkins (cloud-native and on-prem), GitHub Actions, GitLab CI - Track record: 500+ production deployments at Sonrisa (2020-2025) - Also works with: Cloudflare, Proxmox VE, Linux (Arch), Python, Bash, n8n, Datadog, Snyk - Favorite tool: Terragrunt - Interviewing: 70+ technical and hiring manager interviews - Homelab: [planet-express](https://github.com/geripgeri/planet-express), ongoing, with every non-obvious choice written up as an ADR - Smart home: Home Assistant since 2017, 50+ devices, ESPHome fleet, [documentation and translation work for Home Assistant](https://www.home-assistant.io/developers/credits/) - Contact: contact@peidl.net, [GitHub](https://github.com/geripgeri), [LinkedIn](https://linkedin.com/in/gergely-peidl) ## Current work I lead Cloud Infrastructure at Protex AI. I set technical direction, review important pull requests, and write infrastructure code every week. When the Infrastructure Team was formed in May 2026, I decided to step back from people management and returned to a hands-on Lead DevOps Engineer role. ## Leadership outcomes Lead DevOps and Cloud Infrastructure Engineer, May 2026 to present: - I sit at level 4 of 5, the Staff-equivalent band. Protex has no Staff or Principal title, so this is the same scope other companies call Staff - I lead infrastructure incident response and post-mortems, and I am the go-to person for infrastructure issues - I lead AWS cost optimization, rightsizing, and reserved-savings analysis - I am a key contact for a handful of vendors, handling license, extension, and renewal agreements Engineering Manager and Lead DevOps Engineer, Cloud Infrastructure Team, September 2025 to May 2026: - I built a cross-country engineering team from zero, including 1:1s, status cadences, coaching, and team operations - I migrated legacy Terraform to a stack-based IaC pattern with secrets management, Terraform tests, and an ADR program I work best in a player-coach role, combining technical direction, mentoring, and hands-on infrastructure engineering. ## Career history Protex AI, a SaaS scale-up - Lead DevOps and Cloud Infrastructure Engineer: May 2026 to present - Engineering Manager and Lead DevOps Engineer, Cloud Infrastructure Team: September 2025 to May 2026 - Lead DevOps Engineer, Platform Team: July 2025 to August 2025 Sonrisa Informatika, a project-based consulting company - Senior DevOps Engineer and Technical Lead: August 2020 to July 2025, left on Senior II, level 8 of 9 - DevOps Engineer: March 2016 to August 2020 - Software Engineer: October 2013 to March 2016 I mentored multiple engineers at Sonrisa. One of them progressed from entry-level Junior Engineer to Senior Engineer and then to Tech Lead under my mentorship before I left in July 2025. ## Highlights - Star Employee Award, Protex AI Outstanding Contributor Awards, September 2026 - 1st place, AWS GameDay Infinidash Ride-Hailing (AI/ML track), Vienna 2023 - 2nd place, AWS GameDay Chaotic Commencement, Vienna 2023 - 1st place, ISZHOK National Conference of IT Students hackathon, 2013 - Certificates: AWS Certified Cloud Practitioner (2023), IELTS 7.5 / C1 (2024) - Guest lecturer at University of Pécs and University of Pannonia (soft skills for IT careers) - Public speaking: SAFe System Demos to 300+ viewers, conference and company talks ## Notes - This file follows the llms.txt convention: an H1, a summary, then linked sections. Every link is a real URL, on this site or on GitHub. - This site is one static HTML page. No trackers, no analytics. - All profile content is in the served HTML, not rendered by JavaScript, so nothing here is hidden behind a script that a crawler might not run. - Page freshness: the JSON-LD dateModified is bumped by CI on every merged pull request, dated from git history rather than the clock. The sitemap carries no lastmod.