Built from firsthand experience inside the AI industry.

For more than four years, I have worked within the evolving AI industry, contributing to different forms of human-powered AI development and evaluation.

My experience has spanned rating search results and page quality, evaluating news and other content, assessing AI outputs, and working with egocentric video data. As the industry has advanced, I have continued to develop my skills and take on increasingly specialized responsibilities.

Across these different kinds of work, one thing has remained consistent: the importance of the humans behind the systems.

AI capabilities continue to improve, but that progress also depends on people who can interpret instructions, apply domain knowledge, recognize errors, make sound judgments, and produce high-quality examples and evaluations. Their contributions help determine what models learn, how their performance is measured, and where they need to improve.

Working inside these processes has also shaped how I think about talent.

A résumé can communicate experience. A credential can demonstrate formal learning. An interview can reveal how someone communicates. But none of these, on its own, necessarily proves that a person can perform a specific task to the standard a client requires.

The real test is the work itself: Can the person understand the requirements, apply the right expertise, handle difficult cases, and deliver results that meet the standard?

That question became central to my thinking about AI infrastructure.

I began to see an opportunity to build a better way for AI teams to access specialized human expertise: not simply by finding people with relevant backgrounds, but by preparing them for specific work, testing their capabilities, and establishing measurable standards for quality.

That is the idea behind SoReliable.

We are building a system that connects specialized expertise with clearly defined tasks, structured training, rigorous evaluation, and real-world feedback. Our goal is to help AI teams obtain human intelligence they can assess, trust, and scale.

SoReliable is built on a conviction shaped by firsthand experience: reliable AI requires more than powerful models. It requires the right human expertise, applied to the right work, against the right standards.

My responsibility as founder is to prove that this system works: to build it around real customer needs, establish a high bar for quality, and turn that conviction into infrastructure AI teams can depend on.