About New Theory
At New Theory, we're rethinking intelligence from the ground up. Inspired by the mathematical structures observed in nature and emergent in biological and artificial intelligence, we're developing a fundamentally new class of AI architectures. Our models are built with natural inductive biases-like symmetry, compositionality, and hierarchy-enabling radical generalization from limited samples.
We're using geometric deep learning and probabilistic inference to build world models-architectures that capture the structure and dynamics of reality itself. These models are radically more efficient, faster, smaller, and more interpretable than today's transformer-based models, pushing the boundaries of what's possible.
We believe these fundamental innovations enable novel applications in video generation, scene understanding, robotics, and beyond, to grounded reasoning and language.
We're an early-stage company backed by world-class researchers from UC Berkeley and University of Amsterdam, seasoned entrepreneurs, and investors with a bold vision: to make the next step change in AI.
What You'll Do
As a Senior AI Researcher, and an early team member, you'll play a key role in designing and developing the next generation of world models. You'll lead and collaborate on projects that involve:
- Developing and publishing cutting-edge research in geometric deep learning, generative models, and world models, including equivariant networks, graph neural networks, and computational Riemannian geometry.
- Prototyping models that achieve state-of-the-art performance with orders-of-magnitude less data and compute.
- Working closely with our engineering and product teams to apply models to robotics, video generation, and 3D scene understanding. Starting with Video Generation.
- Exploring new mathematical frameworks that challenge conventional Deep Learning paradigms.
Who You Are
- You have a PhD in Deep Learning, Mathematics, Physics, or a related field, or equivalent research experience. Postdoc and/or equivalent industry research experience preferred.
- You've published in top conferences (e.g., NeurIPS, ICLR, ICML, CVPR, ECCV).
- You have deep expertise in geometric deep learning, group theory, differential geometry, graph neural networks, and/or generative modeling at scale
- You're fluent in PyTorch, JAX, or similar, and you write clean, research-grade code, implementing novel architectures from scratch.
- You think like a scientist, and build like an engineer.
- You're energized by big questions and unafraid to rethink core assumptions.
Bonus Points
- Experience with robotics, video synthesis, 3D perception, or physics-based neural simulation.
- Contributions to open-source libraries or research communities.
- Experience mentoring junior researchers or shaping research direction.
- Experience with model performance optimization
- Project management/team lead experience
What We Offer
- A rare chance to help build a transformative AI company from the ground up.
- A collaborative, ambitious, and highly technical team.
- Flexible salary, significant early-stage equity, and benefits.
- A culture of curiosity, creativity, and autonomy-with a fun team on a real mission.
How to Apply
Drop us a line with your CV, a link to your Github and Google Scholar, and a few words on what excites you about building the next wave of intelligence:
And let us know if you have any questions, we're always happy to chat.