About the job
About the Fellowship
This prestigious Postdoctoral Fellowship at Basis Research offers a unique opportunity to collaborate with the renowned Ellis Lab at Cornell University. As a fellow, you will play a vital role in the groundbreaking MARA project, focused on the development of foundational AI technologies. This initiative aims to empower systems to actively discover abstract models of the world and utilize them for effective reasoning to achieve their objectives.
About Basis
Basis is a nonprofit organization dedicated to applied AI research, striving to accomplish two interrelated goals. The first goal is to understand and construct intelligence, establishing the mathematical principles behind reasoning, learning, decision-making, understanding, and explanation, while developing software that embodies these principles. The second goal is to enhance society’s capacity to tackle complex challenges, expanding the scale and complexity of problems we can address today, and accelerating our ability to solve future challenges. To achieve these objectives, we are building a new technological foundation inspired by human reasoning, along with a collaborative organization that prioritizes human values.
About Kevin Ellis’ Group
Under the leadership of Kevin Ellis, Assistant Professor in the Computer Science department at Cornell University, the Ellis Lab conducts research in artificial intelligence, program synthesis, and the intersection of AI and cognitive science. The lab explores methodologies for constructing AI systems that learn and reason akin to human cognition, emphasizing areas such as programming by example, world modeling, neural-symbolic integration, and few-shot learning. By integrating techniques from machine learning, program synthesis, probabilistic programming, and cognitive science, the group aims to develop AI systems capable of learning intricate tasks from minimal data and generalizing across various domains.
Research Focus
Our research is centered on establishing new foundations and technologies for modeling, abstraction, and reasoning within AI systems, with a primary focus on the MARA project. The overarching goal of MARA is to create systems that actively discover abstract representations of the world and utilize these representations to accomplish specific objectives. Achieving this will require significant advancements in knowledge representation, abstraction techniques, and reasoning methodologies.
