Neat Things I Do
Four things I do. One way I do them: honestly, thoroughly, and with code you can read. Whether you're a government agency, a research institution, a foundation, or a development team — the approach is the same. I assess your situation, tell you what I actually think, and build something that works. No vendor lock-in, no black boxes, no surprises.
Free Software for Public Good
Governments, foundations, research institutions, NGOs, collectives, and think tanks deserve infrastructure they can inspect. I help organisations adopt free software — from strategy through migration to long-term support.
What this looks like in practice:
- FOSS adoption strategy and roadmap development
- Migration planning from proprietary to open-source systems
- Vendor-independence audits — identifying lock-in risks
- Long-term support and maintenance agreements
Who this is for: Any organisation that serves the public interest and wants infrastructure it truly owns.
The Right Tool for the Right Job
Not every infrastructure problem deserves the same solution. Puppet for configuration management and compliance — because state enforcement and auditability matter when you're managing a fleet. Ansible for automation and one-off tasks — because not everything needs a persistent agent. Containers for ephemeral, process-heavy workloads — because some applications are built to be disposable.
Every problem deserves its own solution, neatly applied.
What this looks like in practice:
- Puppet module development, deployment, and fleet management
- Ansible playbook design for automation and orchestration
- Container strategy (Docker, Podman) for application workloads
- Bare-metal provisioning and lifecycle management
- Infrastructure testing and compliance validation (InSpec)
- Fixing what's already there — a bad implementation doesn't always need a migration, sometimes it just needs doing right
- Hardware revival — extending the life of existing servers well beyond their accounting write-off. In a market where big tech is buying up all the RAM and GPUs, keeping solid hardware running longer isn't nostalgia — it's economics
Who this is for: Operations teams and infrastructure engineers who need reproducible, auditable, and well-governed infrastructure. Also: anyone whose current setup was built with good intentions but needs someone to untangle the results.
We Watch. We Fix. You Sleep.
If it runs, it should be monitored. I design and implement self-hosted monitoring and observability stacks — built on tools you control, running on infrastructure you own.
What this looks like in practice:
- Zabbix deployment for infrastructure and network monitoring
- Prometheus and Grafana for metrics, dashboards, and alerting
- Log aggregation and analysis pipelines
- Alert design and on-call escalation workflows
- Capacity planning and performance baselines
- Site reliability and observability as a service
- Incident management and analytics reporting — so you know what happened, why it happened, and what we're doing about it
Who this is for: Any organisation that wants visibility into its infrastructure without sending metrics to a third-party SaaS.
Neatly Innovative AI
I believe AI should earn trust, not demand it.
We live in a time where AI is taking away the humanity of humans — generating art nobody asked for, replacing jobs instead of empowering them, optimising engagement over wellbeing. Meanwhile, most people just want AI to handle the chores — the repetitive, the tedious, the mechanical — so they can focus on the work that actually requires a human being.
That's what I build for. AI governance tools that keep humans in control, that are transparent about what the AI did and didn't do, and that earn trust through openness rather than demanding it through lock-in.
Products:
- Rosett-AI (nncc) — Author AI development rules in YAML, compile to Claude Code, Cursor, AGENTS.md, and more. One source of truth for your team's AI governance. GPL-3.0-only. Discover Rosett-AI →
- auditor-mcp — MCP server for InSpec compliance automation. Define controls, run audits, and validate dependencies.
- openvox-mcp — MCP server for OpenVox Puppet ecosystem governance. Attribution, licensing, security, and scope control.
Who this is for: Development teams adopting AI-assisted workflows who want governance they can trust and inspect — and who believe AI should serve humans, not replace them.
How an Engagement Works
1. We talk.
You tell me about your situation. I ask questions — probably more than you expected. No charge for the initial conversation.
2. I assess.
I look at what you have, what you need, and what's realistic given your timeline and budget. This is where the honest counsel starts — if I think you're solving the wrong problem, I'll say so before you've spent a euro.
3. I propose.
A clear scope, a realistic timeline, and a fixed engagement structure. No open-ended retainers that bleed budget. You know what you're getting and what it costs before work begins.
4. I build.
At your table or from mine — but always in direct communication with you. No intermediaries, no handoffs. You see the work as it happens.
5. I hand over.
Everything I build, you own. GPL-3.0-only for software, CC BY-SA 4.0 for documentation. Full knowledge transfer — because the goal is for you to run it without me, not to create a dependency on me.
What you won't get:
- A 40-page slide deck before any real work starts
- Surprise invoices for scope that wasn't agreed
- Deliverables you can't inspect, modify, or redistribute
- A consultant who vanishes after the contract ends