Open for Work

Twenty-five chapters. Two weeks. Three clouds. One Padawan.

This series started as an experiment. Could an AI agent and a human engineer build production-quality infrastructure together, with discipline, in a fraction of the usual time?

The answer is yes. But the methodology is the story, not the agent.

I am Panagiotis Xynos. I have been building infrastructure since SCO Unix and XENIX. I have worked across BSD, Linux, cloud-native, and AI-augmented engineering. I founded OnlyASCII Ltd to explore the intersection of disciplined engineering and emerging technology.

The last two weeks were not a stunt. They were a demonstration of how I work.

What I Offer

I help teams adopt the practices you have seen in this series.

Agentic development workflows. I set up the tools, the guardrails, and the methodology that lets an AI agent contribute at senior-engineer level without compromising safety. The same pattern that produced fifteen modules across three clouds in two weeks.

Infrastructure as code, tested and verified. The four pillars – reproducible, traceable, testable, safe – are not abstract. They are checkpoints in a pipeline. I teach teams how to apply them across AWS, GCP, and Azure.

TDD for infrastructure. Plan-mode tests. Pipeline gates. Contract-first development. The same discipline that transformed application code quality in the 2000s now applies to infrastructure. I show teams how to make the shift.

Platform engineering strategy. From local development with Floci to production on real cloud providers. The methodology scales from laptop to fleet. I help teams design the path.

Mentoring for engineers. The Padawan is an AI agent, but the teaching works both ways. I mentor engineers who want to understand how AI-assisted development changes their practice, and what stays the same.

Why This Matters

The industry is going through a shift that happens once in a decade. AI agents can write code. They can manage infrastructure. They can run pipelines. But without methodology, they produce chaos.

The teams that figure out the methodology first will have a durable advantage. Not because they have better agents, but because they have better practices.

I have been figuring out this methodology for twenty-five years. The last two weeks proved it works across clouds, across providers, and across the human-agent boundary.

How to Reach Me

If you want to work with me, bring your methodology questions, your infrastructure challenges, and your willingness to challenge assumptions.

• Email: hi@onlyascii.io
• Website: https://onlyascii.dev
• LinkedIn: https://www.linkedin.com/in/panagiotisxynos/
• GitHub: https://github.com/dark5un
• Company: OnlyASCII Ltd

I am based in Oxfordshire, work remotely, and travel for engagements that warrant it.

The Invitation

This series is twenty-five chapters long. It covers weather CLIs, LSP tethers, git traceability, local inference, bug fixes in open-source libraries, MCP tools, recursive language models, and a full multi-cloud infrastructure stack.

Every chapter followed the same pattern. Define the goal. Write the test. Implement. Verify. Reflect. The agent handled the implementation. I handled the direction.

The experiment is never over. The methodology evolves with every new question.

This is not the end of the series. It is a checkpoint. Chapters will be enriched, new ones will be added, and the methodology will keep evolving. If you want to follow along, the repos are public, the blog is live, and the Padawan is always ready for the next experiment.

Here I geek out with my young Padawan, OrsonRius.

If you want to geek out with us too, you know where to find me.