My career evolution with AI

With AI, it's been quite a journey. Back in 2023, knowing prompt engineering was enough to feel ahead of the curve. In 2024, building a basic RAG app or chatbot was enough to stand out. In 2025, building AI agents put you in the top ten percent. And in 2026, the bar is moving again. People are now looking for engineers who understand how AI products actually get built: multi-agent workflows, multimodal apps, context engineering and optimization, evals, and how to make these systems work in production. And the best way to really learn this stuff is by taking classes and building things yourself, which is something I've approached pretty seriously.

These days there are a few areas I focus on in my job:

Vibe coding. This is where I spin up prototypes to quickly test ideas with users. A few years ago, this would have meant days in Figma, or some other tool, just to get something clickable in front of people to validate a new feature. Now I can put together a surprisingly convincing flow locally in a matter of hours, hand it to a user, and hear what they think almost right away.

Building my own tools. I've also realized I can build little tools to handle mundane tasks that are very specific to me. Consolidating reports, pulling data from a few different analytics systems into a single dataset and visualizing it, spinning up Jira tickets without all the clicking around, and so on. 

AI-augmented processes for my team, to give everyone a performance boost. Engineering is driving most of this, but I make a point of keeping pace and moving along with them. In practice that looks like AI-assisted code reviews that catch the obvious issues before a human even looks, or auto-generating first drafts of documentation and release notes so nobody starts from a blank page. We've also leaned on it to summarize long incident threads into a tidy postmortem, and to draft test cases straight from a ticket's acceptance criteria.

Learning where AI actually delivers value. And then there's simply taking the time to understand how AI applies to real applications and platforms, the kind that deliver genuine value, whether to customers or internally. That means learning in a classroom setting, connecting with teams who've already done this and seen results, following what's happening in the media, and working through my own use cases.

I also challenged myself and built a full production application end to end. It has a back-end running in Azure, OpenAI calls powering a data transformation workflow, a few APIs exposed to consumers, and then a web app on top with proper user management, API calls, and rich UI elements. I did it partly to play with the tools, partly to actually learn to vibe code, and partly to understand what it takes to get an app from zero to full production with coding agents doing a lot of the heavy lifting. My software background definitely helped, but I'm still impressed by how easy the whole thing turned out to be. Read about my vibe-coded project here.

My favorite online classes are right here. As new developments land in the AI field, fresh classes keep getting added, which makes it easy to stay current. And these are bootcamps, no joke. They take real time and persistence.