Apply on Employer Site

Stability AI · 16 hours ago

Research Scientist – Geometry (AI Assisted)

United States

Full-time

Remote

Senior Level

Stability AI is seeking a Research Scientist with expertise in geometry processing and a passion for modern learning systems. The role involves developing AI-assisted geometry processing pipelines and learning-based representations for 3D geometry, while collaborating with various teams to enhance creative and production workflows.

Artificial Intelligence (AI)Generative AIImage RecognitionInformation TechnologySoftware

Responsibilities

Research and develop AI-assisted geometry processing pipelines for UV unwrapping, remeshing, geometric reconstruction, and shape generation

Design learning-based representations for geometry (meshes, point clouds, implicit fields) that capture structure, topology, and correspondence

Develop token- or patch-based encoders for shape representation, enabling compression, editing, and reconstruction from learned codes

Integrate learned geometry modules into generation and reconstruction frameworks, ensuring geometric validity and multi-view consistency

Build training and evaluation pipelines with quantitative metrics for distortion, reconstruction fidelity, and mesh topology quality

Collaborate with graphics, simulation, and ML teams to bring new geometry models into creative and production pipelines

Contribute to publications, benchmarks, and internal best practices in geometry + AI research

Qualification

Geometry ProcessingMachine LearningPyTorch/JAXDiscrete GeometryDifferential GeometryGenerative Modelling3D WorkflowsResearch SkillsCollaboration Skills

Required

PhD (or equivalent experience) in Computer Graphics, Geometry Processing, Machine Learning, or a related field

Deep understanding of discrete and differential geometry, including remeshing, surface parameterization, and geometric optimization

Experience with learning-based geometry representations (e.g., geometric autoencoders, tokenization, learned unwrapping, or generative reconstruction)

Strong engineering skills: proficiency with PyTorch/JAX, geometry/mesh libraries, and large-scale experiment pipelines

Ability to bridge classical geometry processing with modern learning-based techniques and apply them to practical 3D workflows