Research Scientist – Diffusion Models

Granica•San Francisco, CA

8h•Onsite

About The Position

Diffusion models have transformed image, video, and multimodal AI. We're applying those ideas to one of the next frontiers in machine learning. At Granica, we're building Large Tabular Models (LTMs)—foundation models designed to learn natively from enterprise data. Realizing that vision requires new generative modeling techniques capable of learning from structured information at scale. Our research is led by Prof. Andrea Montanari (Stanford) and explores a fundamental question: How can diffusion models enable the next generation of AI for enterprise data? If you're excited about inventing new generative learning algorithms and applying them to entirely new domains, we'd love to talk.

Requirements

PhD in Machine Learning, Computer Science, Statistics, Applied Mathematics, or a related field.
Strong research record in generative machine learning.
Experience developing new generative models or learning algorithms.
Hands-on experience with PyTorch or JAX.
Strong programming skills in Python.
Ability to turn research ideas into working systems.

Nice To Haves

Experience with diffusion models, score-based generative modeling, representation learning, probabilistic modeling, or scalable ML systems is particularly relevant.
Research applying diffusion models beyond traditional vision tasks.
Publications at NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, or related venues.
Open-source or production ML systems experience.

Responsibilities

Develop novel diffusion models and generative learning algorithms.
Research new representation learning techniques for Large Tabular Models.
Design efficient training methods for large-scale generative models.
Prototype and evaluate new generative modeling approaches.
Design rigorous experiments and benchmarks to measure model quality and efficiency.
Collaborate closely with Prof. Andrea Montanari and Granica's research team to translate research into production systems.