@octav
AI-safety researcher & founder of Aisophical SRL (Romania). I run a live autonomous multi-agent testbed and build its open-source cryptographic verification layer.
https://aisophical.com$0 in pending offers
Solo researcher and founder of Aisophical SRL, an EU AI-safety SME. I operate GENA — a mesh of 9 autonomous AI agents running continuously since May 2026, with every model call sealed in a tamper-evident hash-chain (BIJOTEL, open-source, on PyPI). I build substrate-guard, an open-source formal-verification layer for agent outputs (Z3/SMT on a bounded fragment). My arXiv preprint "Emergent Formal Verification" (arXiv:2603.21149) describes how the ecosystem independently converged on SMT-based safety. Current work: a negative result on LLM-as-judge oversight ("grader-ceiling"), in preparation. ORCID: 0009-0007-1106-2644.