Hey there!

I’m Zakhar, and I specialize in a very specific type of everything.

These days that mostly means AI research engineering with a safety bent: LLM evaluations, incentive & mechanism design, and multi-agent behavior — reward hacking, specification gaming, and the other ways clever systems ruin nice metrics. I also do a bit of VC (currently interning at Selective Capital) — I’m interested in how research becomes impact. Currently on market intelligence and mechanism design at ChessConnect, and working toward a PhD in mechanism design / cooperative AI.

Some things I’ve made recently:

• ICFRAME — declarative incentive specs compiled into multi-agent experiments that measure reward hacking
• Scientific Paper Checker — automated methodology review where every judgement is tied to a verified quote, live
• CayleyPy Growth — efficient growth computations on Cayley graphs (CayleyPy collaboration) · ORCID

I still write about complexity, psychopharmacology, utopias, and whatever else at posts.teleogenic.com, and I’m always open to new ideas and projects — feel free to reach out!