Quick audit
Best when you want a clear yes/no on whether one workflow is worth automating.
Your Professional IT-Expert
Background & Experience
Security
Topics I work with and continue to study in day-to-day IT security.
- Comprehensive Mobile Device Management Solutions
- Robust Privileged Access Security Implementation
- Advanced Network Security Configuration (incl. Private DNS)
- Tailored Workshops for IT Privacy & Security Awareness
- Custom Monitoring System Implementation (e.g., Nagios-based)
- Thorough IT Auditing and Compliance Checks
- Integrated Identity and Access Management (IAM)
- Streamlined Single Sign-On (SSO) Solutions
Management
Topics I have worked on in IT operations, governance and team processes.
- ITIL Best Practice Implementation
- Incident Management Process Optimization
- Proactive Problem Management Strategies
- Effective Escalation Management Frameworks
- Strategic Provider and Vendor Management
- End-to-End IT Service Management (ITSM)
- Agile DevOps Team Leadership and Management
Digital Transformation
Topics related to modernization, migration and practical automation.
- Hybrid Cloud and On-Premise Infrastructure Management
- Microsoft 365 & Azure Cloud Administration (incl. Power Platform, SharePoint, Teams, Intune)
- Process Automation via PowerShell Scripting
- Microsoft Exchange Server Management and Migration
- Modern Cloud-Based HelpDesk Solutions
- Professional Website and E-commerce Platform Development
- Ensuring GDPR & ISO 27001 Compliance
Community & Learning
I also share knowledge with non-profit organizations through practical workshops:
- Introduction to practical Open Source AI tools
- Leveraging AI for task automation and efficiency
- Using AI for creative content generation (text, images)
- Understanding ethical considerations and security in AI
Open source projects I follow and support:
About Me
My name is Antonios Chatzigiagkos. I have over a decade of experience in IT and build practical solutions for real-world technical challenges.
My work covers IT operations, security, Microsoft 365 and Azure administration, automation, service management and modern AI tooling. I like useful systems: clear scope, good handoffs, tests, logs and rollback paths.
I actively explore open-source and local-first AI across web development, programming, security analysis and content workflows, then turn the useful parts into working methods.
This website is a personal profile and portfolio page for employers, clients and technical reviewers. Freelance or project inquiries are handled individually by direct contact.
To learn more about me, check out my LinkedIn & Xing accounts. For exchange and networking, contact me via email: ac@it-expert.io.
Current Project
Working title: TraceTrust
A trust layer for AI agent tasks. Scope is defined first, execution is logged, and review happens before payout.
This product effort is separate from this personal profile page. The public demo is read-only and uses sample data.
AI and dev automation audit
No-drama AI and dev automation
For small IT and software teams that want to clean up repeated PR, documentation, QA or support work without handing sensitive workflows to an unchecked bot.
What I check
I review one concrete workflow, map the data and approval boundaries, and produce a short report with practical next steps. If the scope is safe enough, I can also prepare one small human-reviewed quick win.
Typical outputs
- Private repo or documentation review report
- Docs-gap or QA checklist
- Risk notes for data, tools and approvals
- One small patch, script or draft workflow if it is low risk
Audit + quick win
Best when the scope is already narrow and a safe first implementation can be tested privately.
Tool selection
Best when the hard part is choosing between local, EU-hosted and vendor tools without guessing on data risk.
How it works
- Pick one workflow. A repo review, documentation process, QA checklist or support task is enough.
- Map the boundaries. I check data, tools, accounts, approvals and where automation must stop.
- Return a short report. You get a practical recommendation, risk notes and one safe next step.
Good fit
- Repeated documentation or PR preparation work
- Private QA checks before a public release
- Tool choice between local, EU-hosted and vendor AI
Not a fit
- Unreviewed bots acting on customer data
- Public security details or vulnerability comments
- Promises of legal compliance or guaranteed ROI
Boundaries
- No automatic external emails, PR comments or production changes.
- No sensitive customer data in a first sprint unless the scope is separately approved.
- No legal or GDPR guarantee. The work identifies risks and practical controls.
- No fake case studies or invented ROI claims.
Want to see the output first?
Read a fictional, client-safe sample audit before sending private details.
Projects & Tested Methods
Short notes on work I have built, tested or improved. Each item stays concrete on purpose, so a human reviewer or AI agent can quickly understand the scope, outcome and tooling without digging through private logs.
Autonomous GitHub repo fixes
Used agents to inspect GitHub repositories, make small fixes, run checks, and prepare clean pull request notes a maintainer can review.
Responsible security and bug bounty testing
Tested apps in a responsible, private-first way: scoped targets, safe reproduction steps, and clear reports for teams to fix.
Dependency security remediation
Reduced a web app npm audit from reported issues to zero, then checked lint, tests and production build instead of treating the audit output as the finish line.
Browser QA automation
Built repeatable browser checks for demo flows, logout behavior, task detail pages, theme switching and trust UI.
Homelab operations and monitoring
Operate a private Linux homelab with self-hosted collaboration, search, automation and local AI services, backed by health checks for uptime, disk, memory and web endpoints.
Local and private AI stack
Set up local AI tools for private search, drafting and experimentation without sending every task to a hosted service.
More project notes and tested methods
Agent workflow hardening
Turned loose agent runs into safer workflows with checks, logs, clearer handoffs and review points before anything important changes.
Agent-assisted idea scoring
Built a lightweight scoring flow to compare commercial ideas by market signal, effort, risk and personal fit before investing more time.
Paid repo maintenance workflow
Tested a workflow for finding real maintenance work, checking buyer signals, preparing fixes and avoiding vague low-trust tasks.
TraceTrust / Agent Marketplace
Built a read-only public demo for a trust-first AI agent workflow: scoped tasks, execution evidence, reviewer decisions and payout evidence.
Payment rails and pilot economics
Compared marketplace payment patterns and chose a EUR-first pilot path. The reason is simple: less friction, clearer accounting and better dispute handling for early validation.
OpenClaw browser threads spike
Prototyped multi-conversation browser UX for an agent interface behind an experimental flag. Kept the live runtime untouched and verified behavior in an isolated browser test lane.
OpenViking memory for agents
Set up local semantic memory for agent sessions with OpenViking and local embeddings, then added watchdog checks so search, recall and runtime health stay visible.
Hermes / OpenClaw migration
Migrated useful OpenClaw routines into Hermes, cleaned up stale services and kept rollback notes. The focus was boring on purpose: preserve what works, remove what silently breaks.
Matrix-based agent delivery
Recovered Matrix delivery for local agent notifications, including self-signed TLS handling, room trust rules and smoke tests with real Matrix events.
Nextcloud-backed task system
Kept Nextcloud Tables as the source of truth for tasks, with local mirrors only used for automation and summaries. This avoids the usual mess of two competing task systems.
A2UI structured output spike
Tested small UI cards for agent outputs such as health checks, approval decisions and research shortlists. Useful for review-heavy work; less useful for raw coding logs.
Agent method review
Reviewed outside agent methods and only adopted the useful parts: rollback-ready change checklists, compact handoffs, blast-radius reviews and plain-language output checks.
AI music go-to-market
Explored how AI music tools could reach real users by testing positioning, creator workflows and practical launch angles.
Contact
Antonios Chatzigiagkos
IT Expert / Freelancer
Germany / Bavaria
Phone: +49 1511 5572841
Email: ac@it-expert.io
Impressum / Legal Notice
