Senior AI Engineer · Ex-ING · Full-stack developer
Temporary account. My original is @vladimirrotariu, with contributions to PyTorch, SurrealDB, Opik, and more.
From the mathematics of quantum scattering to AI in production. I work where the math is the hard part: machine learning, quantitative finance, zero-knowledge proofs, and digital twins, with full-stack development around them.
SysKnife: an MCP server and CLI that let an AI agent administer Linux through typed, approval-gated actions. The model picks an action, a policy-enforced daemon builds the command, and every change lands in an Ed25519-signed audit log. Written in Rust.
mojo-dllm: local inference for diffusion language models such as LLaDA and Dream, written in Mojo.
New contributors are welcome on SysKnife and mojo-dllm. The good first issues and help-wanted issues above are scoped so you can finish one without knowing the codebase.
Also mine:
- claude-math: renders math legibly in the Claude Code and Codex terminals, with Unicode glyphs in place of raw LaTeX.
- maintainer-agent: the unattended maintainer that reviews pull requests and keeps the issue tracker on SysKnife. It approves a change only against proof that a guard bites.
- patton-protocol: a hook for Claude Code and Codex that asks before a coding agent changes a file that existed when the session started, tested in pre-registered studies across ten agent harnesses.
Education
- BSc Mathematics, specialization Pure Mathematics, University of Groningen (2017 - 2021)
- BSc Physics, specialization Theoretical Physics, University of Groningen (2016 - 2021)
Links
- 📄 Bohmian insights into mathematical scattering theory, which closed an open problem: whether the Bohmian interpretation of quantum mechanics can be reconciled with experiments.