Unconventional AI
AI Theory & Systems, Algorithms & Models
Mountain View, CA +1 · Full-time · Posted Jul 21, 2026
- Location
- Mountain View, CA
Mountain View - Education
- MS/PhD or equivalent research/project experience in a quantitative field such as AI/Machine Learning, Computer Science, Physics, Electrical Engineering, or Applied Math.
About the job
Design and scale multimodal AI architectures that combine conventional neural networks with novel computational primitives and run efficiently on the company’s computing substrates.
What you'll do
- Co-design, implement, and evaluate AI models and components, including diffusion, flow, energy-based, and continuous-time or implicit architectures.
- Develop hardware-specific scaling laws and methods to improve model capabilities across modalities.
- Translate between theoretical and hardware constraints, coordinating model requirements with systems and hardware teams.
What they ask for
- Deep expertise in the theory, training, and empirical analysis of modern ML architectures, with demonstrated ability to design, implement, and scale models across tasks and modalities.
- Strong understanding of conventional and continuous-time or implicit architectures and their computational trade-offs.
- Deep experience with PyTorch or JAX, including custom model components and training loops.
Nice to have
- Experience with numerical methods for differentiating through continuous-time systems.
- Perspective on co-designing model architectures for unconventional computing paradigms.
Tools
PyTorch, JAX, Transformers, Mixture of Experts, Neural ODEs, Deep Equilibrium Models, diffusion models, flow matching
AI focus
Develop multimodal AI models and architectures—including image generation, language and reasoning, video, and world models—optimized for novel compute substrates.
Model-written summary · the original posting is authoritative