Follow for automation and AI agent patterns that survive production, built in the open and shared without the hype.
Most enterprise automation breaks the moment it meets the real world. I build the kind that does not.
17+ years building software and automation for teams of every size, from solo founders to enterprises. I turn manual, error-prone processes into systems a team can trust. I build agent frameworks and Claude Code tooling, and I share what works so other people can skip the expensive lessons.
The six repositories pinned on this profile.
| Project | Why it is worth a look |
|---|---|
| Download VSIX From Marketplace | Grab any VS Code extension as a VSIX, any version, any OS. Used daily by developers who hit the marketplace wall. |
| App Store compliance | The open playbook for App Store and Google Play rejections, plus a pre-submission guard an AI coding agent can run. |
| AI Workflow Delivery Framework | Professional delivery standards for n8n automation consultants. Open and free to use. |
| hetzner-mcp | The Hetzner platform as tools an AI assistant can call. Cloud, Storage Box, and dedicated servers, cost-guarded. |
| patterns | A software-patterns reference. Eighteen dimensions on every entry, every claim cited. |
| headless-claude | Field-tested patterns for running Claude Code headless without surprises. |
Several platform teams independently picked me to represent their product, and two more brought me into their partner network. That is the part of a resume you cannot write yourself.
OpenAI Partner University. API Expert, ChatGPT Expert, and Codex Expert.
Certified n8n Expert Partner, licence 18628384, valid through February 2028. Code merged into the n8n core repository.
| Focus | What that means for you |
|---|---|
| Automation systems | I design and ship n8n and AI agent systems. Scoped, documented, and built to survive a real audit. |
| Team training | Hands-on Claude Code and n8n workshops. Your team leaves able to build it themselves, not watch from the back row. |
| Building in the open | Tools, frameworks, and patterns I share so you do not have to learn them the hard way. |
- 17+ years building across enterprises including Accenture, Capco, Deutsche Bahn, BMW, and Telekom Deutschland
- Automation work measured in seven-figure operational savings
- Speaker at W-JAX Munich. Host of the Frankfurt and Munich n8n meetups, and of Clay Club Frankfurt.
- Working with teams from solo founders to enterprises, across the DACH region and beyond
- Master's at the University of Passau. CLIMB credential from Harvard Business School.
Automation is the core skill. The agents, the clouds, and the models sit on top of it. The public half is the table above. The rest lives inside client work, and the lessons from both end up in what I share.
Automation is the mastery. Give me the constraint. I have shipped the edge, the cloud, the data, the product, and the API between them. The stack is whatever is already in production. I am not attached to one vendor.
The agents sit on this machine. Each one has a job.
| On this machine | How I use it |
|---|---|
| Claude Code | The coding agent I hand a task to. Headless when the job should finish without me watching. |
| Codex | A second pass on the same change, before it ships. |
| Grok | The long research thread, before the code starts. |
| n8n | The core skill, in a tool. Orchestration a team keeps after the project ends. |
| Models | OpenAI and Anthropic on this machine. DeepSeek, Qwen, Zhipu GLM, Kimi, and MiniMax when the job, the price, or the data residency says so. The other major Chinese labs speak the same API, so they land in the same workflow. |
| Mastra | TypeScript agents, when the agent is the product. |
| Clay | The data behind outbound. Enrichment and research, then n8n carries it into the CRM. |
| v0 | The first interface. Then I make it something a team can ship. |
| APIs | If a system speaks, I can put it in the workflow. REST, webhooks, MuleSoft, and the ERP that only documents half of its endpoints. |
| MCP | The way an agent reaches a live system. I publish the servers. Hetzner and Xentral are on npm. |
| npm | Community nodes under nex8n. Xentral, ClickFunnels, DocuSign, Suno, and a PII redactor. |
| Automation | |
| Models | |
| Go to market | |
| Cloud | |
| Runtime | |
| Product | |
| Data | |
| Delivery |
n8n for orchestration. Claude Code, Mastra, and the Anthropic stack for agents. TypeScript, Next.js, and React on the web. Python, Rust, or Swift for the product. Cloudflare in front. AWS, Azure, Google Cloud, or Hetzner behind it, including the bank migrations. Postgres, MySQL, Redis, or BigQuery where the data has to last. Linux, Docker, Kubernetes, and Terraform when it has to scale. The piece in the middle is an API, an MCP server, or an n8n node. I have shipped all three. The model is a choice. OpenAI, Grok, Gemini, Mistral, Llama, DeepSeek, Qwen, Alibaba, Zhipu GLM, Kimi, MiniMax, or the next lab that speaks the same API. Outbound runs through Clay, Instantly, Apollo, HeyReach, and Smartlead. Voice is ElevenLabs. The CRM, the ads platforms, and the site sit on the same workflow.
Updated November 2026.
Mirza Iqbal, in Frankfurt and Munich. Official n8n Ambassador and Expert Partner. The work is the automation a team keeps after the project ends.
Yes. datev-mcp lets an assistant read bookkeeping, payroll, and documents, and it previews a write before anything is sent. What people asked for this past month is a real DATEV import and a booking proposal, not another bookkeeping product.
The cluster this past month is a missing privacy manifest, Play testing and verification, and a payments-policy link. The open playbook is app-store-compliance.
No. Cloudflare in front when it is public. AWS, Azure, or Google Cloud when the estate is already there. Self-hosted n8n when the data should stay on a machine you already own, including a Mac mini, or on Hetzner.
Whichever the job needs. OpenAI, Grok, Gemini, Mistral, Llama, DeepSeek, Qwen, Alibaba, Zhipu GLM, Kimi, MiniMax, or the next lab that speaks the same API.
AI moves faster than any team can staff for, and most of the noise is not worth chasing. If your team is buried in manual steps, or working out where automation and AI genuinely fit, here is the short path.
- Start free. The AI Workflow Delivery Framework above is open and yours to use today.
- Learn it. Claude Code and n8n workshops, on-site or remote. Book a session.
- Go further. If your team wants to move faster with automation, that is what I do at next8n.
next8n.com | Book a session | LinkedIn | X
Build the thing. Ship it. Then teach someone else to.