I build AI systems that solve real business problems, usually at the intersection of product, engineering, and operations.

I co-founded OBK LLC, where we build AI agent systems and workflow automation for enterprises and SMBs. Before that, I spent 4 years as a founding engineer at TradeWizz (an Amazon FBA SaaS) and co-founded Albatros Caravan, an IoT hardware startup. Both taught me more through failing than through succeeding. I'm currently finishing my MSc at IYTE, researching generative models for turbulent flow reconstruction. On the side, I like building small AI tools and experiments whenever something catches my curiosity.

• AI agent systems that take over repetitive business processes, from customer support to content production to order handling
• Process transformation with AI-powered automation, helping companies restructure their workflows before plugging in AI
• MSc thesis comparing GAN and diffusion models for turbulent flow reconstruction at IYTE

• The hardest part of an AI project is not the model. It is getting people and processes ready for it. If the organization is not prepared, the best system will sit unused.
• Most complexity problems are structure problems. If you can organize the chaos, the solution usually becomes obvious.
• AI agents work best when they handle specific, well-defined, repetitive tasks. The hype usually skips this part.
• Adoption decides whether a project succeeds. Technology alone does not.
• I learn best by building. If I want to understand something, I make a small version of it.

Some client names are kept confidential. Click any project for the full case study.

Small tools I build when something catches my attention. More on the way.