From Building in Azure to '96 Code: Why Legacy Made Me a Better AI Engineer

The first team: building from scratch

My first job was as a Software Engineer on a well-structured team: agile Scrum with a DevOps mindset, working closely alongside architects, PMs, and QMs. We built services from scratch in Azure while maintaining the ERP core.

I learned a ton—integrations, system design, testing. But what impacted me most was when the architect started coming to me to ask how I did things. He loved how I prepared diagrams and often asked me to walk him through them. As a junior, that validation was huge. It made me feel like my contributions truly mattered.

The shock: legacy, absence, and the AI wave

The shift to legacy arrived in 2026, bringing a major change of context. We took on projects that were unfamiliar territory: legacy code, decisions made back in '96, and the classic challenge of these systems: decades of accumulated knowledge silos and high technical complexity. Overnight, the environment we had mastered completely shifted.

Even though I had more experience, understood the product better, and grasped both the technical core and business needs, finding room to demonstrate value in that environment was tough. When feedback loops fade, self-doubt creeps in fast: do I really know as much as I thought?

I haven't lost ability. I've changed context. But that feeling is hard to shake when expectations are high.

In the middle of all this, I went on 17 weeks of parental leave. I cherished every second with my daughter, but the industry didn't pause. While I was away, both the market and the company accelerated their AI adoption initiatives. To top it off, I received news of a promotion while still on leave—a new title on paper, but with recognition that didn't quite match the jump in responsibility or the moment.

When I returned, the shock was real. Not only was it hard to reconnect with a project that felt distant, but I also had to get up to speed on my own to catch up with technology that had advanced rapidly while I was away.

Between the legacy systems, the tech gap, and the feeling of half-baked recognition, I felt like legacy too. Unsure of my trajectory, it sometimes felt like everything was moving too fast and I was constantly playing catch-up.

Rebuilding: where I am now

For a long time, I waited for that validation to come back from the outside. It didn't. What actually worked was to stop waiting and start building.

I came to realize that dealing with legacy software wasn't a step backward—it was a reality check: AI and modern software don't live in a pristine sandbox; they have to integrate and solve friction right where code hurts the most.

Today, I'm still at Unit4 on the same team—an incredible group hardened by tackling complex technical debt and taking over tough projects. But I knew the shift in perspective had to come from me.

I set the frustration aside and started looking for real pain points to solve: from automating daily team tasks with Python and PowerShell scripts to reviving the creative drive that led me to build Reinterpreter for our internal innovation event.

Above all, I shifted my focus toward AI. Today, I build and integrate intelligent tools not because it's trendy, but because it forces me to learn just like I did on my first team: fast, pragmatically, unafraid of making mistakes, and solving real day-to-day needs.

Validation didn't return through external applause or a line on a pay stub. It returned when I started feeling that what I build serves a real purpose.

What I take with me

If I learned anything from that first team, it's that I love building things. I enjoy continuous learning and solving complex problems. And I know confidence doesn't disappear; sometimes it just gets buried under an unsupportive context.

Navigating legacy code didn't hold me back—it gave me the context required to become a more mature engineer with a grounded perspective.

I'm writing this as a reminder to my future self. That feeling of impact and mastery exists: it might get temporarily buried under technical debt and life transitions, but you can always dig it back up with the right mindset and tools.