Research
Ruth works on specification inference — the question of how to construct formal specifications from observed agent traces, so that future runs can be checked against specifications nobody has had to write by hand. The work shares tooling with the verification framework and the mechanistic-interpretability stack; Ruth is one of the contributors who works comfortably across all three.
She co-authored the debate-plus-trace result and is one of the named contributors to the verifiable-policies paper. Her Ph.D. was on relational specifications in distributed systems; the lab's current specification work is recognisably an extension of that.
Ruth is based in Vienna. She is one of the most active reviewers on the cross-axis methodology review pool and has a habit of asking, on first read of a new methodology, 'what is the smallest counter-example.'
Background
Ph.D. mathematical logic, TU Vienna, 2015.
Prior to alphabell: TU Vienna; Cantor Initiative; Wayfarer Institute.
Selected publications
-
Sep 2025 · ab-scalable-oversScalable Oversight for Multi-Step Agent Systems: a Debate-Plus-Trace ApproachIfeoma Nwosu-Howard, Hiroshi Tanigawa, Maral Lotfi, Ruth Wernicke
-
Mar 2025 · ab-verifiable-polToward Formal Verification of Learned Policies in Bounded EnvironmentsAviva Stern, Sun Kyung-min, Felipe Avelar
-
Sep 2024 · ab-interpretabiliInterpretability Cell Pairing: how every dual-use capability run gets a watchful siblingKarima Belkadi, Hester Vandekerckhove, Yuki Cho
-
May 2025 · ab-mechanistic-ciMechanistic Circuit Analysis at Frontier Scale: cells as a unit of interpretabilityJiang Yifei, Nico Almgren, Karima Belkadi, Hester Vandekerckhove
-
Nov 2024 · ab-sandboxed-selfSandboxed Self-Modification: a confinement specification and implementationLiora Sabatini, Cheung Wai-Lin, Marek Holub
Recent talks
- Specifications from traces, FM 2025
- What pairing reveals about specifications, ML Safety Workshop 2024
Ruth is currently part of node-cell lebesgue-22, working under the Interpretability & alignment research axis. The cell is open to substantive correspondence from researchers working on adjacent problems; route requests through lebesgue-22@alphabell.com or directly to Ruth at ruth-wernicke@alphabell.com.
Contact
- EMAIL
ruth-wernicke@alphabell.com - ORCID
0000-5476-7871-2890 - X
@ruthwernicke - BLUESKY
ruth-wernicke.bsky.social - GITHUB
@ruthwernicke
Cross-references