Aviva Stern

Research

Aviva is the lead author of the lab's verification framework paper (25/01) — the formal combination of abstract interpretation of the policy network and symbolic execution of the environment dynamics that proves safety properties of deployed cells' policies in bounded environments. The result is one of the more technically demanding pieces in the interpretability axis.

She is currently leading the cell-internal effort to extend the framework to a wider class of environments. The work is incremental — each weakening of the boundedness requirement is its own self-contained result — and Aviva publishes it that way.

Aviva is based in Tel Aviv and works closely with the regional academic cluster on the formal-methods side of the framework.

Background

Ph.D. mathematics, Hebrew University of Jerusalem, 2014.

Prior to alphabell: HUJI; Cantor Initiative; Helios Safety Group.

Selected publications

• Mar 2025 · ab-verifiable-pol
  Toward Formal Verification of Learned Policies in Bounded Environments
  Aviva Stern, Sun Kyung-min, Felipe Avelar
• Sep 2024 · ab-interpretabili
  Interpretability Cell Pairing: how every dual-use capability run gets a watchful sibling
  Karima Belkadi, Hester Vandekerckhove, Yuki Cho
• Sep 2025 · ab-scalable-overs
  Scalable Oversight for Multi-Step Agent Systems: a Debate-Plus-Trace Approach
  Ifeoma Nwosu-Howard, Hiroshi Tanigawa, Maral Lotfi, Ruth Wernicke
• May 2025 · ab-mechanistic-ci
  Mechanistic Circuit Analysis at Frontier Scale: cells as a unit of interpretability
  Jiang Yifei, Nico Almgren, Karima Belkadi, Hester Vandekerckhove
• Nov 2024 · ab-sandboxed-self
  Sandboxed Self-Modification: a confinement specification and implementation
  Liora Sabatini, Cheung Wai-Lin, Marek Holub

Full publications index →

Recent talks

• What 'bounded' actually buys you, CAV 2025
• From policy nets to proofs, POPL 2024