Michael Page is looking for an Enterprise Architect (m/w/d) for its client.
Let’s take a look under the hood. Our patient today is a German provider of legal information from Saarbrücken. The company was founded in 1985 and is partly state-owned. What does this mean for us, the IT professionals? It means almost 40 years of IT history, layers of technologies, government standards, and most likely a very mature (read: old) and complex system landscape. This is not a startup in a garage; everything here is serious and tangled.
And now, they are looking for an “Enterprise Architect.” If we translate from corporate speak to human, it’s a cry for help. They need someone to come in and sort out this zoo of technologies accumulated over decades. The tasks in the job description speak for themselves: “modernization, harmonization, and optimization of the system landscape,” “implementation of architectural principles,” “creation of architectural blueprints,” “management of domain maps.” Essentially, they need a Moses to lead their IT through the desert of legacy to the promised land of microservices and clouds. A titanic task, requiring not only intellect but also nerves of steel to combat bureaucracy and entrenched habits.
But what if I told you that this role could be filled not by a weary veteran of IT battles with a history of burnout, but by an indefatigable digital assistant? What if this headache could be outsourced to artificial intelligence? Sounds provocative? Perhaps. But let’s break it down.
Imagine not a person, but a system. Let’s call it, for argument’s sake, the “Architectural Oracle.” Instead of spending months interviewing teams, trying to understand how one system connects to another, we do the following.
Step 1: Digital Archaeology. We “feed” the AI absolutely everything we have: code repositories (Git), server logs, data from monitoring systems (Prometheus, Grafana), configuration files (Terraform, Ansible), Jira tickets, all available documentation in Confluence, even if it hasn’t been updated since the dinosaurs. The AI, unlike a human, won’t get tired or miss a detail. It will analyze millions of lines of code and logs to build an up-to-date, living map of your IT infrastructure. Not pretty Visio diagrams that are outdated the moment they’re saved, but a dynamic model reflecting reality.
Step 2: Strategic Modeling. With this map, the AI can do what would take a human weeks. It can identify bottlenecks, find duplicate functions across different services, highlight security risks, and pinpoint areas with excessive technical debt. You can ask it questions: “Oracle, show me all services that still use the old authentication library.” Or: “Model what would happen if we moved this monolithic module into a separate microservice. How would that affect performance and cloud costs?” The AI will calculate dozens of scenarios and present you with an analysis backed by numbers, not subjective estimates.
Step 3: Automated Governance and Oversight. New “architectural principles” are great. But how do you get everyone to adhere to them? Instead of a human architect walking around teams and banging their fist on the table, the AI assistant is integrated directly into the CI/CD pipeline.
Imagine a dialogue:
– “Listen, Hans,” says a young developer to the veteran architect, “I’m trying to deploy a new service, but the system won’t let me. It says I’m using direct database access, bypassing our new API gateway. This AI guardian is a devil!”
– “And that’s a good thing, Jürgen,” Hans replies, sipping his coffee. “Before, I would have found out about your rogue activity three months later, when everything crashed. Now, the system catches it in five minutes. It’s my best assistant. It frees up my time to think about strategy, not chase every commit.”
To reduce distrust, there’s no need to jump in headfirst. Start with one team or one domain. Let the AI assistant work in “advisor mode,” simply highlighting problems, while a human makes the decisions. When teams see that the tool genuinely helps find errors and save time, they will come asking for more authority for it themselves.
How to validate the results? It’s simple. Business metrics. We implemented the “Oracle.” What changed after six months?
– Did the time-to-market for new features decrease?
– Did the number of critical incidents in production decrease?
– Did cloud infrastructure costs decrease due to optimization?
If the answers are “yes,” then our digital architect is doing its job. And the human whom Michael Page is so diligently searching for could become not an executor, but an operator of this system. Someone who asks the AI the right questions. But that’s a completely different, far less stressful, and more high-level job opening, isn’t it?
Источник: https://www.linkedin.com/jobs/view/4406660223/
