I’m Peter Skliros—a technical consultant and platform engineer with 15 years across cloud, identity, endpoint management, automation, and enterprise change. I work comfortably from whiteboard to production.
Deepening the overlap between dependable infrastructure, useful automation, and carefully supervised AI agents.
Four examples across enterprise delivery and hands-on experimentation. Open a file for the short case study.
Fifteen years moving between discovery, architecture, hands-on implementation, stakeholder guidance, and the unglamorous final mile that makes change stick.
Translate ambiguity into a practical roadmap. Make the trade-offs visible. Bring engineers, vendors, and decision-makers into the same conversation. Then stay close enough to delivery to know the design survives reality.
I’m useful where complex technology, real operational pressure, and human communication meet.
Desktop automation becomes risky the moment an agent can click, type, launch applications, and alter files. I wanted a safe lab for exploring agent-driven workflows without handing an experiment the keys to my everyday machine.
Python • VM isolation • Desktop control experiments • Guardrails and deliberately limited permissions
Where agents genuinely remove friction, where they become unpredictable, and which permissions, checkpoints, and recovery paths make them safe enough to be useful.
I explore new tools with curiosity and operational discipline in equal measure. If I cannot explain the blast radius, it belongs in a disposable VM.
ClickOps doesn't scale, and neither does doing the same portal task forty times.
PowerShell • Entra ID • Intune • AWS
Turn known-good operational steps into readable, reviewable scripts with clear inputs, useful output, and safe failure behaviour.
Less repetitive administration, fewer opportunities for drift, and a process that another engineer can understand and run.
I don’t automate for theatre. I automate work that should become boring, predictable, and easy to hand over.
Salt water is an unforgiving runtime environment. Parameters drift, corals file incident reports by dying expensively, and manual test kits don't scale past enthusiasm.
Controller • pH / temp / ORP / salinity probes • dosing pumps • automated RO/DI water production • dashboard/alerting
Continuous parameter monitoring, dosing, alerts, and automated RO/DI water production. The desktop widget streams representative data; the corals retain their privacy.
The same instincts carry across domains: observe the system, understand failure modes, automate carefully, and keep a human-readable escape hatch.
The first version of this place proved I could build a convincing little world. The next version has a harder job: show how I think, what I can do, and where I’m heading—without sanding off the personality. The serious work now gets the front door. The arcade moved into After Hours. The seal refused all proposed changes.
Spent the weekend wrestling with some PowerShell to automate a tedious infrastructure task. There's a very specific kind of joy that comes from spending six hours writing a script to save yourself ten minutes of manual clicking. It's about the principle, really. At least, that's what I tell myself after the third cup of coffee.
Every site needs one. This is mine. More to come once I finish configuring the host environment.
Hand-picked links, no algorithm involved.
Games and odd experiments live here now. They show craft and curiosity without making you fight through an arcade to find the professional work.
Platform Engineer · Technical Consultant · Systems Thinker
● ACTIVE BUILD · OPEN TO THE RIGHT NEXT CHAPTERIt's now safe to turn off
your computer.