I lead engineering teams. I still write code.

I'm Alex Janjic, a VP of Engineering based in Canada. I lead the teams building AI and software systems, and I stay close to the implementation.

What I do

Leadership means setting direction, helping engineers grow and deciding what's worth building. I care about the decisions that connect an engineering team's work to a useful, reliable system.

Hands-on work means writing code, testing ideas and investigating the details: why something is slow, why an agent made the wrong decision or what happens when a dependency fails.

That connection between leadership and implementation keeps my decisions grounded. Building things helps me see the constraints. Leading teams makes me think beyond whether the code works.

What I've worked on

I've improved business workflows by integrating AI into the way the work gets done, rather than adding it as a separate tool.

I've built a unified internal AI platform connecting business systems and workflows. I've introduced autonomous software development workflows and used AI-assisted engineering to improve delivery throughput.

I've also reduced infrastructure and operational costs through AI-assisted analysis and engineering improvements, and brought AI capabilities into customer-facing enterprise products.

That work spans the systems themselves and the way teams build, release and operate them.

Why Context Window exists

When I'm learning something, I want to understand the decisions behind it. What did someone try first? Why did they change their approach? What looked promising but failed once they tested it?

Context Window is where I share those details: working code, measured experiments, architecture decisions and the limits that matter when a system has to run. I distinguish experiments from deployed systems: a benchmark tests a particular question; it isn't proof that an approach will hold up in production. Failures, unknown outcomes and the boundaries of each result belong in the explanation, not outside it.

I also explore what AI-native organizations need beyond individual tools: how responsibility, human judgment and operating practices fit together. The AI operating model is a way to examine those choices, not a claim that every organization should work the same way. It's written for engineers building AI systems and the leaders responsible for them.

The goal is to make the next person's starting point a little better than mine was. When an approach has limits, I'll explain them. When I get something wrong, I'll correct it.