αlphabell.
α · People · Hester Vandekerckhove
Stylised avatar of Hester Vandekerckhove

Hester Vandekerckhove

Long-tenured contributor

Based in Brussels
Node-cell hilbert-13
ORCID 0000-3943-7967-4726

Research

Hester works on the operational backbone of the interpretability axis. Her contributions to the paired-cell protocol have been mostly invisible — the kind of work that turns 'pair an interpretability cell with the RSI cell' into a concrete sequence of artefact subscriptions, signed pairing records, and disagreement-handling state machines that two cells can actually run for six months.

She is one of the small number of contributors who has been a paired interpretability lead on a halt that was called — twice, in fact, on agentic-axis sandboxed self-modification experiments in 2024. She has written about both halts in the internal index.

Hester is based in Brussels and contributes to the European Union's AI Office advisory pool on a personal basis. She is careful, in her writing, to distinguish lab positions from her personal positions; she would like that to be the lab's general practice.

Background

Ph.D. mathematics, KU Leuven, 2015. Postdoc at Cambridge DAMTP, 2015-2017.

Prior to alphabell: KU Leuven; Praxis AI Studies; Wayfarer Institute.

Selected publications

Full publications index →

Recent talks

  • The pairing protocol as a small contract, ICLR 2025
  • Disagreements that produced halts, ML Safety Workshop 2024
Working with

Hester is currently part of node-cell hilbert-13, working under the Interpretability & alignment research axis. The cell is open to substantive correspondence from researchers working on adjacent problems; route requests through hilbert-13@alphabell.com or directly to Hester at hester-vandekerckhove@alphabell.com.