> whoami
AI-Native Engineering Leader
I build engineering teams and ship applications where AI agents are first-class architectural components. That means agent workflows, tool-integrated delivery, governance, and team adoption. 28 years of production systems across FinTech, Advertising, News Media, and eCommerce.
> cat ./pivot
For the past 18 months, I have been singularly focused on the AI transformation of software engineering — not as a side interest, but as a full professional pivot. I have immersed myself in LLM capabilities, agentic architectures, prompt engineering, MCP integrations, multi-model orchestration, and AI-native development workflows. I've shipped AI-first policies and tooling in production, built open-source developer tools for the AI ecosystem, and published practitioner-level writing on what AI-native engineering actually looks like when it's not a demo. 28 years of systems thinking didn't become obsolete — it became the foundation.
> ls ./links
> cat ./why-hire-me
28 years of architecture and delivery depth — from CNN in 1996 to leading a distributed FinTech engineering team today.
Proven AI-native transition inside a real engineering org — AI-first policies, MCP integrations, agentic workflows, and a structured bootcamp shipped in production.
Built open-source agent tooling and orchestration systems — Orkestr, Skillr, and Tessera: from agent runtime to prompt management to trust-aware context retrieval.
> cat ./case-study
AI-Native Transformation — FinTech Engineering Team
Problem: A distributed engineering team (Germany, Montenegro, India, Latin America) was shipping features the traditional way — manual code reviews, manual issue triage, no structured approach to AI tooling. Developers were experimenting individually with AI assistants but without shared practices, governance, or measurable impact.
What I changed:
- Established an AI-first code generation policy as the team-wide standard for all new feature work
- Integrated MCP servers with self-hosted GitLab to automate issue triage and merge request analysis
- Deployed AI agents for project management workflows and automated code review
- Designed and delivered a structured AI-native bootcamp teaching a five-role separation model (Think / Execute / Edit / Review / Iterate)
Result: The team moved from ad-hoc AI experimentation to a repeatable, governed AI-native workflow. Issue delivery and code review turnaround cut by at least 50%. The five-role model became the shared vocabulary for how the team works with AI tools — not as a novelty, but as infrastructure.
> cat ./experience
Head of Software Development & Systems Architect
Immotege GmbH · Leonberg, Germany (Remote) · Nov 2020 – Present
Led architecture, development and AI transformation of a multi-tenant FinTech SaaS platform.
Senior Back-End Developer
ITPlusX GmbH · Weilimdorf, Germany · Jul 2016 – Sep 2020
Built RESTful APIs and web applications for diverse agency clients. Led modernization of legacy systems to microservices.
Senior Back-End Developer
Werbewelt AG · Stuttgart, Germany · Apr 2015 – Jun 2016
Built Laravel-backed solutions for SPAs and eCommerce. Introduced modern Git workflows.
PHP Developer / Infrastructure Engineer
boingMedia GmbH · Göppingen, Germany · Jul 2014 – May 2015
Modernized internal tooling and managed Linux infrastructure.
Freelance Web / PHP Developer
Self-Employed · Atlanta, GA, USA · Feb 2010 – Dec 2013
Delivered CMS and eCommerce solutions for small businesses.
Web Developer / Lead Engineer / Senior QA Automation Engineer
Various (CNN, mediaOcean, VeriSign, IHG) · Atlanta, GA, USA · Jan 1996 – Jan 2010
Early web development pioneer at CNN. Led development and QA teams across major media and tech companies.
> ls ./projects
Orkestr
Self-hosted Agent OS for designing, executing, and governing autonomous AI agents on your own infrastructure. Open-source (MIT). Three layers: Orchestration (visual DAG workflow builder, canvas, schedules, checkpoints), Agents (agent loop with MCP tool calls, A2A delegation, persistent memory, 18 pre-built agents), and Components (reusable skills, multi-model access, version history, provider export). Provider-agnostic — mix cloud and local models, run fully air-gapped with Ollama.
PHP 8.4 · Laravel 12 · React 19 · TypeScript
Skillr
Open-source universal AI skill & prompt manager. Write skills once in portable YAML + Markdown, sync native config files to Claude, Cursor, Copilot, Windsurf, Cline, and OpenAI with one click. Features multi-model test runner (Anthropic, OpenAI, Gemini, Ollama), version history with diff viewer, skill composition, template variables, prompt linting, and a marketplace with 25 pre-built skills.
Laravel 12 · Filament 3 · React 19 · TypeScript
Tessera
Mac-first personal context vault with policy-gated semantic retrieval — the trust substrate for agentic AI. Curate what matters, then grant AI agents scoped access with full auditability through disclosure receipts. Default-deny architecture: nothing automatic, minimize disclosure, prove what happened. SwiftUI desktop app with Rust backend (Axum gateway, SQLite vault, ONNX embeddings).
Rust · SwiftUI · macOS · Axum
AI-Native Development Bootcamp
Open-source three-session curriculum (90 minutes each) that moves engineering teams from AI-assisted to AI-native development by assigning the right tool to the right role. Teaches the five-role separation model (Think/Execute/Edit/Review/Iterate), an orchestration decision matrix, tool capability boundaries, and team conventions for responsible AI-augmented delivery. Stack-agnostic, MIT-licensed, with worked adaptations for Node/TypeScript, Python/Django, Go, and Rails. The curriculum behind the 'Upgrade Path Available' essay — shipped in production inside a real FinTech engineering org.
AI-Native Development · Team Training · Curriculum · Five-Role Model
AI-Native Engineering Doctrine
Portable, fork-and-curate scaffold of AI-native engineering doctrine — standards, agent skills, tool adapters, and review rubrics designed to travel between projects, interviews, contracts, and roles. Doctrine-only (no runtime): the layer that informs what goes into a harness like Claude Code, Cursor, or a self-hosted agent OS — which CLAUDE.md content, which skills, which permission boundaries, which review rubrics. Eight curated sections from engineering canon to security and review rubrics, with Claude as the reference tool adapter and lighter sketches for Cursor, Copilot, Codex, and a generic AGENTS.md. MIT-licensed and public-safe.
AI-Native Engineering · Engineering Standards · Coding Standards · Agent Skills
> ./contact --reason
Hiring for a senior architecture or engineering leadership role?
Need a founder-friendly advisor for agent workflows or AI-native delivery?