/ DevSecOps consulting · Tallinn
Infrastructure you can trust. AI with guardrails.
Fixed-price AWS audits, hardened Kubernetes and AI-agent guardrails, from an engineer who has run production since 1996.
01 How I work
Three rules, applied to every engagement.
Evidence, not guesses
Facts are separated from hypotheses. A conclusion exists only after a check that could have proved it wrong.
Read-only first
Look before touching. Nothing changes in your systems until you have seen the plan and the blast radius.
Guardrails, not hope
A fix is not done until the whole class of error can no longer fail silently.
02 Services
Three ways to work together.
AWS audit
Cost, security, infrastructure as code and blast radius in one read-only review. Fixed price, written report.
Details and pricesHighly available Kubernetes
Design, build or review hardened HA clusters, on-premise or in the cloud: etcd quorum, networking, GitOps, secrets, backup and recovery.
Discuss a projectAI guardrails review
Where AI agents may read, what they may change and who approves it. Permissions, scope limits and approval gates for AI in your CI/CD and operations.
Discuss a review03 Tools
Open source, built for the work above.
InfraLens
Read-only AWS audit tool: what runs, what it costs, what looks unused, what depends on what. Produces a client-ready report. One binary, runs on your machine with your credentials.
View on GitHubaiheal
Self-healing CI with AI triage and a human approval gate. The AI may edit only infrastructure files.
View on GitHub04 Book
Don’t Trust. Don’t Fear. Don’t Beg.
Infrastructure Engineering with AI.
AI will accelerate your engineering, or burn your production cluster in forty-five minutes. The outcome depends on the guardrails you build around it.
23 chapters, 490 pages. A hands-on build of a hardened high-availability Kubernetes platform, where every chapter dissects a real outage, and an AI agent works inside limits you define.
05 About
Vakhtang Mosidze. In IT since 1996.
I started in 1996 as the IT person at a hospital, then ran IT at an airline, built a cybersecurity company for banks and oil and gas, and went on to CTO and head of DevOps roles. I have built a 1,500-node compute cluster and moved production to Kubernetes. Physicist by training. These days I help European teams see what is really running in their infrastructure, and keep it under control. Based in Tallinn.
Not sure where to start?
Twenty minutes, no slides, no obligation.