I'm Aleksei Papushin — a DevOps architect who builds and runs production systems, and trains engineering teams to use AI coding agents where it actually counts: in production, under real constraints.
It usually starts the same way — a team that wants AI in its engineering workflow, and someone with 10+ years of running production systems who has actually shipped with it, not just talked about it.
Hands-on programs that take your engineers from cautious to fluent with Claude Code and AI coding agents — on your repos, your stack, your review standards. Where agents earn their keep, where they don't, and how humans stay in control of merged code.
For leaders deciding how AI fits the delivery pipeline. We design the guardrails and team habits that make agentic workflows safe to adopt at scale — so the productivity gain survives contact with your codebase.
Architecture and review for systems running real workloads: infrastructure, delivery pipelines, reliability and cost. Pragmatic, reversible, cheap to operate.
By email or Telegram, free and with no commitment. You tell me what your team is trying to do; we see whether there's a fit.
I look at your code, workflow or infrastructure and come back with scope, timeline and a fixed proposal.
Training or hands-on work with your engineers, with regular check-ins and no surprises.
Documentation and a clean handover, so your team owns the result — with support afterwards.
Sessions in English. Remote across Portugal and Europe, on-site in Portugal.
The teams that win with AI aren't the ones that prompt the hardest. They're the ones who know exactly what to delegate, what to review, and what to never hand over.
I moved to Portugal in 2023 — from long grey winters to the ocean, mild winters and a sun that actually shows up. My family and I have made it home: we bought a house here, and we are staying.
Portugal has remarkable engineers and a tech scene that's growing fast. What it doesn't always have is early access to the newest ways of working — the tools and habits teams in San Francisco or London picked up a year ago. That's the gap I want to close: helping Portuguese teams adopt AI in engineering now, not five years from now.
One honest note: my Portuguese is still at the bom dia, obrigado stage. I work in English — as most of the tech scene here already does.
I teach AI-assisted engineering because I use it to ship. Here's the work that backs the advice.
Cloud cost control for dev and staging. I founded it and designed its architecture: a Python platform where teams lease AWS and GCP machines for a fixed window over Slack, Telegram, or web — the VM auto-stops when the lease ends, and every shutdown stays a human decision.
Visit idlefy.comHardened developer VMs on AWS, built with Terraform and CINC: locked down on first boot, re-converged every 30 minutes, with per-VM AWS credentials and an audit trail in Grafana Loki. Setup is guided by Claude Code skills, and Idlefy switches the fleet off when nobody is using it.
View on GitHubOne Helm chart to deploy any application to Kubernetes: deployments, jobs, cron jobs, services, ingress and Gateway API routes from a short values file — with Prometheus monitoring, cert-manager TLS and werf support built in.
View on GitHubNo. Sessions and communication are in English, as in most Portuguese tech teams.
Remote across Portugal and Europe, and on-site anywhere in Portugal.
Engineering teams that already ship software and want to use AI coding agents seriously — from startups to established product companies.
Since February 2025, Article 4 of the EU AI Act requires companies to ensure AI literacy among staff who work with AI systems. Hands-on training is a practical way to address it.
I sign an NDA, and your code stays in your environment, under your access controls.
Mostly Claude Code, but the practices are tool-agnostic and carry over to other coding agents.
Every engagement is scoped after our first exchange and the diagnosis. You get a fixed proposal before any work starts.
As a registered sole trader in Portugal (ENI), with a Portuguese NIF.
Training, architecture, or a straight second opinion on your AI strategy. Tell me what your team is trying to do — I'll tell you honestly whether and how I can help.
