Julian Hilzinger
Freelance software architect & AI engineer
I bring AI agents safely into grown IT systems.
“Outstanding in implementing complex requirements.”
Three starting points companies hire me for.
I modernize monolithic legacy systems incrementally into scalable microservice architectures, following the strangler-fig pattern and domain-driven design, without big-bang risk. In a three-year engagement I extracted 20+ modules from an enterprise codebase this way.
MCP servers, agent orchestration, and the governance layer around them: verification loops check every result, guard-rails in CI/CD stop misbehavior, cost controls cap consumption. That keeps autonomous workflows auditable and EU-AI-Act-compliant.
Team enablement for agentic software development: I take engineering teams from Level 1 (autocomplete) up to Level 5, as far as their tests and pipelines can carry it. Each level adds the checks it needs.
Team enablement in detail →The model I use to place teams — and the level I work at myself. The jumps to 4 and 5 are an architecture question, not a tooling one.
Where most teams stand today. Both speed up typing; neither changes the process.
An agent works directly in the repository: it reads code, edits files, and runs tests, per task and with human approval of every step.
Agents implement entire features spec-driven: specification → plan → implementation → verification.
Multi-agent systems plan, implement, and verify on their own; the human sets goals and owns the outcomes.
No result counts until a test or another agent has checked it. Only what I approve reaches the main branch. The process in detail →
Software vendor
AgriTech Company
Public Sector Client
1st Place
Connect Claude directly to this site and query profile, services, career and coverage area.
https://hilzinger.dev/api/mcpclaude mcp add hilzinger --transport http https://hilzinger.dev/api/mcpclaude.ai: Settings → Connectors → “Add custom connector” → paste the URL
Read-only by design: the server exposes reading tools only, following the same guard-rail principles I build into production agent systems.
I have been building software for over 10 years, specializing in the modernization of complex legacy systems, cloud-native architectures, and the integration of AI agents. I build the governance layers that let agents do real work without losing track of cost or traceability.
No big bang. I rebuild systems step by step, and they can ship at any point along the way.
Checked before it counts. What an agent delivers only counts after tests and review.
Software architecture & AI engineering, cloud-native solutions, technical lead and digital transformation strategy for clients in industry and the public sector.
Module Mobile Systems and Cloud — teaching iOS, Android, Azure Cloud, software architecture and digitalization.
App development (Apple, Android, web), AR/VR, machine learning, prototyping of new IT technologies and technical lead of a large-scale project (> 800 person-days).
Model Context Protocol: Advanced Topics · Building with the Claude API
Azure AI Fundamentals · Azure Data Fundamentals
Certificate of Completion
EU AI ACT – Compliance & Application
Digital Transformation Strategies
Leveraging Data as a Competitive Advantage
Executive program of the German National Academy of Science and Engineering
Business Strategy, Entrepreneurship, Marketing & Innovation
Best graduate of the M.Sc. Business Analytics program, awarded by the Philipp-Matthäus-Hahn Foundation.
Winner of the Telekom Challenge with AR/VR project. Programming competition with more than 1,300 applicants.
Innovative. Digital. Experience. Award for a digital innovation project at Gühring IT.
Tell me about your project. A first conversation comes with no obligation.
I look forward to your inquiry and typically respond within 24 hours.
How I work: The AI Coding Levels model →