cv

Basics

Name Sam Earle
Label Postdoctoral Researcher at Basis Research Institute; Visiting Researcher at Cornell University
Email sam.e@basis.ai
Url https://smearle.github.io/
Summary Working jointly with Zenna Tavares and Kevin Ellis on intrinsic motivation and procedural environment generation.

Work

  • 2026 - 2026
    Instructor, Game Design
    New York University
  • 2026 - Present
    Visiting Researcher
    Cornell University
    Working jointly with Kevin Ellis and Zenna Tavares on intrinsic motivation and procedural environment generation.
  • 2026 - Present
    Postdoctoral Researcher
    Basis Research Institute
    Working jointly with Zenna Tavares and Kevin Ellis on intrinsic motivation and procedural environment generation.
  • 2025.08 - 2025.12
    PhD Research Intern
    Sakana AI (Tokyo, Japan)
    Exploring open-ended VLM-driven content generation loops under the supervision of Sebastian Risi.
  • 2025 - 2025
    Instructor, AI and Games
    New York University
    Course design, lecturing and project mentorship.
  • 2024.10 - 2025.05
    Research Consultant (part-time, remote)
    Microsoft Research
    Continued work with Andrzej Banburski-Fahey to develop DreamGarden (see below), and explore multi-modal input for automatic game and environment generation.
  • 2024.06 - 2024.09
    PhD Research Intern
    Microsoft Research
    Working under the supervision of Andrzej Banburski-Fahey, as part of the Deep Learning and Speaking the World into Existence teams, to develop DreamGarden, a design assistant for generating simulations in Unreal Engine 5, with a hierarchical planning module and dynamic orchestration between submodules for generating C++ code, 3D assets, and procedural meshes.
  • 2022.09 - 2023.03
    PhD Research Intern
    Facebook AI Research (FAIR) (London, UK)
    Working under the supervision of Roberta Raileanu, as part of the Reinforcement Learning team, to explore the use of pretrained multi-modal generative models for environment generation. Developed DreamCraft, a text-guided NeRF approach for generating structures/environments in MineCraft, with a novel quantization and neural rendering scheme allowing for optimization over both in-game appearance of artifacts as well as their functional properties (i.e. adjacency constraints and block type frequencies).

Volunteer

  • 2008 - 2012

    Ecuador, India, Haiti, Ghana

    Volunteer, Ambassador, and Speaker
    Free the Children
    Free the Children (now "We" charity) is a Canadian non-profit that aims to empower youth by increasing access to education in developing countries. I helped create a club at my middle school dedicated to fundraising and awareness for the charity. During my time on Degrassi, we partnered with the organization and travelled to various countries to learn about and assist with their projects (e.g. building schools, gardens, playgrounds, systems for providing clean water). We acted as public ambassadors for the charity (the trips were filmed and turned into short documentaries that aired on MTV), and I spoke at the charity's "We Day" events in Vancouver, Montreal, Ottawa, and Hamilton.

Education

  • 2020.09 - 2026.05

    Brooklyn, NY, USA

    PhD
    New York University
    Computer Science
  • 2011.09 - 2017.06

    Toronto, ON, Canada

    Honours Bachelor of Science with High Distinction
    University of Toronto
    Cognitive Science, Computational Cognition stream (Major), Computer Science (Minor), Cinema Studies (Minor)

Awards

  • 2026
    Outstanding Reviewer Award
    Genetic and Evolutionary Computation Conference (GECCO)
  • 2025
    Best Paper Award Nominee
    AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE)
    Video Game Level Design as a Multi-Agent Reinforcement Learning Problem
  • 2024
    Best Paper Award
    Foundations of Digital Games (FDG)
    DreamCraft: Text-Guided Generation of Functional 3D Environments in Minecraft

Skills

Programming languages & libraries
JAX
PyTorch
Python
C++
Javascript
Java

Languages

English
Native speaker
French
Fluent