Aziz Berkay Yesilyurt

aziz@appvisor.dev

Hi, I'm Aziz. I build robust software for deep tech. Over the years, I’ve worked with founders and teams turning advanced technologies into real products. Smart decisions early on prevent wasted time, money, and headaches as projects evolve. That’s where good engineering really pays off.

I like to keep things lean and flexible. My work is built to last, easy to change, and focused on what actually matters. People I work with often say I'm fast, reliable, and easy to work with. I move comfortably between design, code, and infrastructure as needed, and I make sure everyone stays on the same page, every step of the way.

I take on new projects occasionally, especially if you’re building at the intersection of science and engineering. If that sounds like you, feel free to reach out at aziz@appvisor.dev.


Featured Projects

Here’s a quick look at what I’ve been working on lately. For more examples, you can browse a few other projects.

Cadlet – Simple 3D Design with Code

Funded by the Dutch R&D Program

Most 3D design tools are hard to automate and even harder to reuse. Designs get trapped in files that don’t play well with code or other systems. I’m building Cadlet to fix that. The idea is simple: describe 3D models with structured text, so they can be shared, automated, and built upon without manual rework.

On-Device AI for Sail Image Analysis

Contracted by Alive Industries & North Sails

I built a cross-platform mobile system that uses on-device AI to segment images and extract sail shape metrics in real time, with no internet connection required. This is a core capability for the product, enabling accurate performance analysis even in offline environments.

RV Campaign – Lead Generation from Walkthrough Videos

Contracted by RV Video Library

RV dealers had great walkthrough videos but no way to use them to generate leads. I developed a system that inserts short campaign clips into existing walkthrough videos without any editing, re-uploading, or technical effort required from the dealers. It runs at scale: over 10 million views per month for under $10 in compute.