Dmytro Hlukh

About Me

I’m an IT specialist based in Switzerland with over 25 years of experience. My focus is infrastructure, cloud migration, automation, and security. I’ve worked in regulated environments across finance, healthcare, and the public sector, often bridging legacy systems and cloud platforms.

I prioritise practical, secure, and maintainable solutions. For recurring tasks, I use AI-supported tools only for drafts, structuring, and first-pass analysis of logs or advisories. AI is never used for autonomous production actions or security-critical decisions. Decisions, prioritisation, and production changes always stay with humans inside clear roles, approvals, and controls.

Selected Projects

Migrated banking services to Azure and AWS with zero downtime and full compliance.
Developed PowerShell automation for Active Directory, reducing manual effort by 70%.
Improved monitoring and alerting for healthcare platforms, increasing issue response by 50%.

Skills

Cloud platforms: Azure, AWS, Oracle Cloud (OCI)
Automation: PowerShell, Bash, CI/CD pipelines
Security: IAM, firewalls, system hardening, zero-trust architecture
Monitoring: Prometheus, Grafana, Zabbix
Reliability: incident management, SLO/SLI, postmortems, and structured handovers
Regulated environments: finance, healthcare, and public sector, with auditability and documented changes
AI workflows: reusable skills, versioned knowledge packs, tool/MCP integration, and human review
Infrastructure: VMware, Windows Server, Linux

Controlled AI workflows

From raw input to reviewable work steps

This is not about autonomous agents doing everything. It is a bounded workflow: a skill defines the task, a knowledge pack supplies approved context, tools stay narrowly scoped, and a human keeps the final decision point.

01InputAdvisory, alert, or rough note

02WorkflowSkill, knowledge pack, and scoped tools

03ReviewHuman checks, approves, or rejects

Drafting, not autopilotAI prepares proposals, but never makes silent production changes.

Limited data scopesOnly approved sources, sanitised examples, and narrow tool access are included.

Versioned knowledge packsRunbooks, SOPs, and checklists stay traceable, auditable, and easy to update.

Human approvalPriority, risk, and execution are always confirmed by people.

Security advisory triage for SMEs
Runbook drafts from rough notes
Structured incident handovers
Infrastructure change review
Explainable output from security and support tools

Loading workflow example...

Skills and Guardrails Practical Example in AI-OS Discuss a Workflow

SME tools

Practical tools for SMEs: the Security check reviews only public TLS and security-header posture and performs no intrusive scans, no load tests, and no active interventions. Security warnings aggregate official advisories (vendors/CERT/NVD) with filtering and export, with no proprietary auto-scoring and no automatic actions. Remote support helps with pre-triage. Radar and Radio Map are visualisation and navigation tools based on official sources and do not run hidden scans. Confidential customer data is not sent unchecked to external AI services.

Security check Security warnings Remote support Open Radar Open Radio Map

See links and references →

Contact

Email: info@hlukh.ch

LinkedIn: linkedin.com/in/hlukh

Video CV: hlukh.ch/vcv-en