A nonprofit applied research organization building universal reasoning engines grounded in probabilistic programming and causal inference to advance society's ability to solve intractable scientific and societal problems.
A nonprofit applied research organization building universal reasoning engines grounded in probabilistic programming and causal inference to advance society's ability to solve intractable scientific and societal problems.
People
Updated 05/18/26Co-founder and Director
Co-founder and Director
Co-founder and Director
Expanding AI Research
Funding Details
Updated 05/18/26- Annual Budget
- $2,946,241
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- $9,633,503
Org Details
Updated 05/18/26Basis Research Institute is a 501(c)(3) nonprofit applied research organization founded in 2022 by Zenna Tavares, Eli Bingham, and Emily Mackevicius. The institute pursues two mutually reinforcing goals: understanding and building intelligence by establishing the mathematical principles of what it means to reason, learn, and make decisions; and advancing society's ability to solve intractable problems by expanding the scale, complexity, and breadth of problems that can be addressed computationally. The organization's approach is grounded in the belief that fundamental principles of reasoning confer a general ability to solve problems, rather than requiring increasingly complex domain-specific systems. Basis draws from advances in programming languages, compilers, databases, and deep learning to build its reasoning technology. The team has deep roots in probabilistic programming, with co-founder Eli Bingham being a co-creator of Pyro, a widely-used probabilistic programming language originally developed at Uber AI Labs. Basis organizes its work around carefully curated challenge problems that require novel advances in reasoning technology while providing real-world impact. Current projects include ChiRho, an experimental language for causal reasoning that bridges modern probabilistic programming with causal inference; R-ADA (Rational Automated Design Agent), which combines large language models with CAD/CAM environments for robot design optimization; participatory city modeling that creates digital city representations incorporating community knowledge; collaborative intelligent systems research studying collaboration across species and ecosystems; and cell and tissue dynamics modeling for biological research. The institute operates as an open-source research organization, distributing its technology freely to domain experts, external researchers, and the broader open-source community. The team includes approximately 28 researchers, engineers, and operations staff spanning expertise in machine learning, neuroscience, programming languages, robotics, and systems design. The organization maintains offices in New York City and Cambridge, MA, and has an advisory board that includes MIT professors Joshua Tenenbaum and Armando Solar-Lezama, Cornell professor Kevin Ellis, Broad Institute CDO Anthony Philippakis, and neuroscience professor Rui Costa. Basis has grown significantly since its founding, with annual revenue reaching approximately $4.7 million in fiscal year 2025. The organization has received funding from sources including Founders Pledge and operates as a subrecipient under a DARPA contract. As of early 2026, the institute is actively hiring across research, engineering, and operations roles.
Theory of Change
Updated 05/18/26Basis believes that building fundamental mathematical principles of intelligence -- rather than scaling up existing approaches -- will create a universal reasoning engine capable of solving problems across domains. By encoding discovered principles into open-source software that serves as a foundation for tackling progressively harder challenges, the organization aims to compound its capabilities over time. Each solved challenge problem produces generalizable knowledge and software tools that make previously unsolvable problems tractable. By distributing this technology as open-source software, Basis seeks to democratize advanced reasoning capabilities and enable researchers, scientists, and policymakers worldwide to address complex scientific and societal problems including causal reasoning in epidemiology, automated scientific discovery, engineering design optimization, and participatory urban planning.
Grants Received
Updated 05/18/26Projects
Updated 05/18/26ChiRho is an experimental language and Pyro extension for probabilistic causal reasoning that lets researchers express causal assumptions as probabilistic programs and causal questions as program transformations, supporting interventions, counterfactuals, causal explanations, and structure discovery.
MARA is a long-term challenge project to build AI systems capable of everyday scientific discovery—proposing hypotheses, running experiments, and constructing abstract world models through active experimentation and reasoning.
R-ADA is a project to automate robot design by treating morphology and control as a sequential design process, combining language models that operate over CAD/CAM tools, simulation-based evaluation, probabilistic modeling, and Bayesian updating to explore and refine innovative robot designs.
Discussion
No comments yet. Be the first to share your thoughts.