αlphabell.
α · People · Maral Lotfi
Stylised avatar of Maral Lotfi

Maral Lotfi

Long-tenured contributor

Based in Toronto
Node-cell lebesgue-22
ORCID 0000-3856-9688-6703

Research

Maral works on formal verification of learned policies — proving safety properties of deployed cells' policies in bounded environments. She co-authored the lab's verification framework paper (25/01) and is leading the cell-internal effort to weaken the boundedness requirement so that the verification approach applies to a wider class of practical environments.

She has been with alphabell since 2019. Her work is one of the clearer examples of the interpretability axis's structurally cross-cutting role: the verification framework is used by world-models cells whose environments fit the formal class, by agentic-axis cells whose policies are simple enough to verify, and by external partners who consume it through the open-source release.

Maral is based in Toronto. She presents the verification framework at non-AI venues — POPL, CAV, OOPSLA — as part of a deliberate effort to keep the work legible to the formal-methods community.

Background

Ph.D. computer science, University of Toronto, 2014. M.Sc. at Sharif University of Technology.

Prior to alphabell: University of Toronto; Halcyon Safety; Cantor Initiative.

Selected publications

Full publications index →

Recent talks

  • Verifying learned policies — what bounded means, POPL 2025
  • Oversight for warehouse policies, RSS 2024
Working with

Maral 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 Maral at maral-lotfi@alphabell.com.