Yakov Shkolnikov

AI Safety & Security · Architecturally Assured AI · Safety-Critical Systems

I build production AI systems, transform how organizations adopt them, and make it safe to experiment. Then I write about what I learn, including what doesn't work.

I founded Soundness AI in February 2026. About me.

Latest writing

Jev and Three Types of Decisions

·LinkedIn·5 min read

A test of TypeSafe's Jev decision model against the game-theory benchmark from How to Fake a Theory of Mind. Jev is near-perfect on dominant-strategy games but largely ignores the opponent's payoffs, and scores 30% on second-order false-belief stories while reporting over 70% confidence. It is fast and cheap for scoring options, but it is not a trustworthy decision system.

More writing All 48 pieces

Recent papers All research and publications