Paul Lichtenberger · DevOps & Platform Engineer

12 years of climbing the stack.

I build and run production infrastructure end-to-end — from CI/CD and containers to identity, observability and LLM operations.

Scope of ownership, 2014 → now

Scroll sideways for the whole profile →

5: companies · automotive → consulting
15+: containerized services in production
0: click-ops · everything as code

Email me Read the CV LinkedIn GitHub Buy me a coffee

location: Austria · remote for the DACH region
languages: German (native) · English (professional) · Spanish (A2–B1)

About

I come from industrial software delivery — automotive, home-appliance and machinery contexts (Continental, EDAG, BSH, Kiefel), where correctness and repeatable pipelines are not optional.

Today I work as a DevOps Engineer at talsen team GmbH, an agile consultancy blending BDD/TDD, DevOps and generative AI in industrial software development. I own the platform end-to-end: provisioning, CI/CD, identity, registry, observability and LLM operations.

My approach is everything-as-code with a strong bias toward EU data sovereignty — I self-host what many teams outsource, so I understand the whole stack down to the metal. I run what I recommend.

Outside of work: cycling in the Alps.

Experience

Five companies, one direction: from writing a single service to owning the platform it runs on.

Feb 2025 – Present

talsen team GmbH current

DevOps Engineer · Remote · Salzburg / Germany

Agile software consultancy blending BDD/TDD, DevOps and generative AI in industrial software development. I own the self-hosted platform end-to-end — containerized services, automation, AI systems and observability.

Design and operate multi-service Docker/Compose platforms (Traefik, Keycloak, SearxNG, Firecrawl, Jitsi, vector databases) with a focus on reliability, secure routing and observability.
Automate provisioning, backup, restore and health checks in Python/Bash, improving resilience and cutting manual effort.
Build RAG ingestion pipelines and chatbots with LlamaIndex, Qdrant/Milvus and Chainlit, plus distributed scraping via Playwright.
Integrate Grafana, Prometheus and logging pipelines across self-hosted stacks for alerting, uptime tracking and fast troubleshooting.
Develop internal developer tooling: VS Code extension logic, Git author automation and CI testing frameworks (Jest + Cucumber).

Docker / Compose
Traefik
Keycloak
RAG · LlamaIndex
Grafana / Prometheus

Jan 2023 – Feb 2025

KIEFEL GmbH

Senior DevOps Engineer · Freilassing, Bavaria, DE

Thermoforming and industrial-machinery manufacturer. Senior DevOps role spanning CI/CD, cloud and the digital infrastructure behind industrial systems.

Built CI/CD pipelines and shaped the architectural design of the digital infrastructure.
Handled cloud configuration and deployment on Azure.
Installed and provisioned industrial PCs with Puppet.
Developed administration tools for the digital infrastructure.

CI/CD
Azure
Puppet
Infrastructure

Jan 2021 – Dec 2022

BSH Home Appliances Group

Continuous Integration Engineer · Traunreut, Bavaria, DE

One of the largest home-appliance manufacturers in Europe. Built the CI/CD backbone and standardized, containerized developer environments.

Developed CI/CD pipelines and rolled out DevOps tools and processes.
Created container images for CI/CD and as a standardized developer environment.
Configured and maintained Jenkins masters and developed shared Jenkins libraries.

Jenkins
CI/CD
Docker
Shared libraries

Apr 2017 – Dec 2020

EDAG Group

Software Developer — Tools (Embedded) · Regensburg, DE

Engineering services provider for the automotive industry. Built developer toolchains and CI/CD for automotive ECU series development.

Developed the Python/MATLAB toolchain for engineers working on series development of automotive ECUs.
Configured and maintained build and license servers; created and supplied Docker images.
Built and maintained CI/CD pipelines with Jenkins / GitLab CI.
Developed PREEvision metrics in Java and supported developers and testers.

Python / MATLAB
Jenkins · GitLab CI
Docker
Java

Jul 2014 – Apr 2017

Continental

Working Student — Backend Development · Regensburg, DE

Global automotive technology company. Backend development for a vehicle remote-diagnosis service, alongside studies.

Developed a Django backend and data model for a vehicle remote-diagnosis service.
Configured and maintained a Couchbase database.

Python
Django
Couchbase

2015 – 2017

Universität Regensburg

M.Sc. Medieninformatik (Media Informatics)

Skills

What I actually reach for, grouped the way I use it rather than as a flat tag cloud.

Platform & Infrastructure

Bare metal to orchestration, everything provisioned as code.

Ansible (end-to-end provisioning)
Docker & Docker Compose
Terraform
Harbor
Kubernetes
Puppet (industrial PC provisioning)
Azure (cloud config & deployment)
SemaphoreUI
Traefik (routing & middleware)
Tailscale (incl. Funnel)

Identity, Security & Registry

Single sign-on, sovereign registries and security tooling.

Keycloak (OIDC / SSO)
Authentik (OIDC / SSO)
Azure AD / Entra ID
oauth2-proxy
traefik-oidc

CI/CD & Quality Engineering

Pipelines and quality gates that hold industrial delivery to standard.

GitLab CI/CD (templates, runners, self-hosted)
Jenkins (masters, shared libraries)
BDD (behave, Cucumber)
TDD (Jest, pytest)
Mutation testing (Stryker, pymute)
Static analysis & complexity tooling
Internal CLI & VS Code tooling

AI / LLM Operations

Running generative-AI workloads privately, with EU data sovereignty in mind.

Langfuse observability
RAG Systems
Vector databases (Qdrant / Milvus)
Chainlit chat UIs
Playwright distributed scraping
Agent deployment & MCP tool servers
Agent safety hooks
Self-hosted OCR / model serving

Observability & Ops

Production monitoring and a broad self-hosted service portfolio.

Grafana
Prometheus (metrics & alerting)
InfluxDB
Promtail / log pipelines
Nextcloud · Matrix/Synapse · Jitsi
SearxNG · Firecrawl
Production service operations

Writing

Field notes from running production infrastructure and self-hosted AI.

Get in touch

Delivery platforms, tightening CI/CD, or running generative-AI workloads privately are the problems I like to talk about.