There are AI “gurus,” and then there’s Alex Pollock: a builder who sweats the details and sweeps away the hype. He’s the kind of person who will question the assumptions everyone else takes for granted, all while keeping an eye on how tech actually lands in people’s hands. If you’ve ever stumbled through a so-called “intelligent” chatbot that missed the point, you’ve felt the gap he’s working to close.

Start with user empathy. Pollock is fond of saying, “A problem well-understood is a problem half-solved.” It’s not just about algorithms and neural nets. Conversations with stakeholders are the backbone of his process. He digs into pain points, hunting for the true drivers behind the request. Imagine a sales team stuck in spreadsheet purgatory, or a scheduling system crashing by lunchtime—he’ll talk to everyone involved, turning their complaints into a real map of needs rather than a wish list.

Instead of diving into code straight away, Pollock prefers paper, whiteboards, and often, lunchroom debates. Once, he overheard a client grumble about data entry during a coffee break. That passing comment led to a simple automation that saved the company hundreds of hours.

When it’s time to architect the solution, reproducibility takes center stage. Alex is notorious for documenting every decision—sometimes to the point of driving his teams a bit mad. But here’s where it pays off: Six months down the line, when someone asks why a model chose route A instead of B, there’s a clear answer.

Let’s talk technology. Pollock’s stack isn’t flashy for the sake of being flashy. Python, TensorFlow, PyTorch sit next to sturdy classics like SQL. He stays skeptical of buzzword ingredients unless they solve a genuine bottleneck. Recently, his team sidestepped a time-consuming image labeling task using off-the-shelf transfer learning techniques, reducing what would have taken weeks into a few afternoons.