Toronto · AI engineering · Open source

I build systems that make intelligence accountable.

I’m Tugrul Guner, an AI engineering leader and builder. My work turns ambitious AI ideas into systems people can understand, operate, and trust.

Read my story Explore ModePot

01 The story

AI becomes useful when its freedom has a shape.

The most interesting engineering problem is not making a model say more. It is deciding exactly where software should be deterministic, where an agent may decide, and how people stay in control.

I lead AI engineering at CDAI and build open-source tools in the gaps I keep encountering: interfaces that should translate cleanly, APIs that need bounded judgment, runtimes that should adapt without complicating application code, and simulations where change must remain replayable.

That work became ModePot—a family of projects built around one idea: powerful systems become easier to trust when their contracts, authority, and execution are visible.

02 Selected work

Four experiments in making software clearer.

01 dexpot Adaptive synchronous Python APIs for GIL and free-threaded runtimes. 02 intpot One typed tool definition, served across CLI, HTTP, and MCP. 03 summonpot Exact operations and bounded agent-owned decisions behind one contract. 04 LifePot An artificial-life world where agent proposals become replayable rules.

03 Resume

A career built around turning uncertainty into working systems.

This page is ready for the complete chronology. For now, the verified professional record lives on LinkedIn and the working record lives in open source.

Current role: AI Engineering Manager, CDAI
Based in: Toronto, Ontario, Canada
Focus: Enterprise AI, applied ML, developer infrastructure
Open source: Creator of the ModePot project family

View LinkedIn View GitHub

The next chapter

Have a difficult AI system to make real?

Start a conversation on LinkedIn