Private cloud & virtualisation
OpenStack and Proxmox VE end to end: design, deployment, operations, major upgrades. Public-cloud exit, disaster-recovery site, backups with tested restores.
I design, build and run servers, networks, private clouds and applications — and I stay responsible for them after handover. 20+ years in production, independent practice in Novi Sad, Serbia since 2022.
OpenStack and Proxmox VE end to end: design, deployment, operations, major upgrades. Public-cloud exit, disaster-recovery site, backups with tested restores.
MikroTik, Cisco, HPE, UniFi, pfSense. VLAN architecture, routing, firewalling, public /24 networks. Farm build-out from scratch: server, edge router, core switch, rack switches.
WireGuard, IPsec, OpenVPN, ZeroTier, VLESS/Reality. Branch offices linked over ordinary internet, remote access for staff, user accounting via RADIUS, resilience to network blocking.
I take ownership of your servers and network: monitoring (Zabbix, Prometheus, Grafana, Loki), patching, verified backups, incident response under SLA, root-cause review afterwards.
Assessment against NIS2 and Serbia's Information Security Law, system hardening, fail2ban, access management, clean-up of compromised websites, audit documentation.
Terraform, Ansible, Docker, CI/CD (GitLab CI, GitHub Actions, Jenkins, TeamCity), AWS. Manual routine becomes version-controlled code. Also: Asterisk/FreePBX telephony and MDM for Mac fleets.
Websites, web services and customer portals, iOS and Android apps, APIs and integrations, Telegram bots, dashboards. You get a working product, not just code — hosting, backups, monitoring and support from the same person who built it. Automated tests and code review are standard, not an upsell.
My own build in Java 17: 8 modules, WebSocket feeds from three exchanges, layered risk controls, shadow A/B testing of strategies, 246 automated tests, daily Telegram reports. It shows how I write code: with tests for race conditions and a post-mortem for every incident.
Production, DR and two staging clusters. Control-plane rebuild with no downtime and no data loss, memory and Galera tuning. A major upgrade was rehearsed first on a 6-node lab, so every trap was found before production.
A new node goes live with one command. Centralised logs and metrics (Grafana Alloy, Loki, Prometheus). Root causes of memory leaks and boot-time failures found and fixed fleet-wide.
15 ZeroTier + MikroTik links and 20+ WireGuard + MikroTik links at seasonal peak, Asterisk in two instances. Analogue telephony fully replaced, no leased lines needed.
Central management through FleetDM and Apple Business Manager: zero-touch enrolment, controlled macOS updates, unwanted software removed from every workstation.
Sole administrator for 20 users on 2 sites: Windows, macOS, Hyper-V, Linux hosting, Asterisk, UniFi and video surveillance, across a seven-hour time difference.
From engineer to CIO: a department of 20 (17 developers, 3 sysadmins), 120+ client organisations, 300+ workstations and servers. Backups with automated freshness checks.
Included without using your hours: monitoring, patching, backup checks, SLA incident response, reporting.
Counted against the 20 hours: deployments, migrations, changes, new users and sites, advisory.
Billed separately: new build-outs, hardware, licences, work outside your coverage window. Unused hours do not roll over.
Invoices in RSD, EUR or USD · payment within 5–10 days · on-site within 100 km of Novi Sad · remote across EU and US time zones
Mon–Fri 9:00–18:00
Outage (P1): 30 min in hours
Degradation / other: 2–4 h / next business day
Reports: —
Mon–Sun 8:00–22:00
Outage (P1): 30 min in window, 2 h outside
Degradation / other: 2–4 h / next business day
Reports: monthly
Around the clock
Outage (P1): 15 min, escalation after 30 min
Degradation / other: 2–4 h / next business day
Reports: monthly + quarterly review
Message me on Telegram or email; I reply within one business day.
I look at your current setup, documentation and 3–6 month plans.
You know the number before work starts, not after.
With documentation, credentials and a rollback plan. Then SLA support or handover to your team.
I started as a system administrator in 2004 and over 17 years grew into the CIO role, leading a department of 20. As a DevOps engineer at a US start-up I reworked CloudFront caching and cut load and latency by 30%. Since 2022 I have lived in Novi Sad and run my own practice.
Everything is documented and you hold the credentials. The Critical tier includes escalation to a second engineer.
Yes — most clients are in the EU, UK and US. Contract and invoice come from a registered Serbian sole proprietor.
Yes. Code, credentials and deployment instructions are handed over in full.
English and Russian.
Describe it in two sentences — I reply within one business day.
© 2026 Shutoff Studio