αlphabell.
α · People · Lin Hao
Stylised avatar of Lin Hao

Lin Hao

Long-tenured contributor

Based in Singapore
Node-cell bessel-04
ORCID 0000-0715-9325-9578

Research

Lin is the principal author of the 70k-environment study and the contributor who most clearly articulates the lab's bet on the unification of perception and prediction. The 70k-env paper began as a question — what would happen if three cells pooled their simulation infrastructure into a single training pool — and turned into one of the world-models axis's most cited results.

He leads bessel-04, the cell that owns the embodied simulation pool. The cell's working mode is unusual: each contributor maintains a portion of the pool, and the cell is structurally tolerant of having different contributors focused on different physics/social/symbolic environments at any given moment.

Lin works primarily from Singapore. He is the lab's primary contact for several Asia-Pacific universities that contribute environments to the pool.

Background

Ph.D. computer science, NUS, 2014. Postdoc at OpenAI (2014-2016, when OpenAI was smaller).

Prior to alphabell: NUS; Volterra Cognition; Wayfarer Institute.

Selected publications

Full publications index →

Recent talks

  • 70k environments, one objective, NeurIPS 2024
  • What we learned by pooling simulators, RLDM 2024
Working with

Lin is currently part of node-cell bessel-04, working under the World models research axis. The cell is open to substantive correspondence from researchers working on adjacent problems; route requests through bessel-04@alphabell.com or directly to Lin at lin-hao@alphabell.com.