Research Scientist Intern, Generative AI (PhD) (multiple locations)

Meta; Zürich; On-site; CHF 22,000 - 40,000; Full time

Job summary

Meta is seeking an exceptional PhD student to explore generative approaches for computer vision and graphics, focusing on 3D/4D scene reconstruction, understanding, and generation. You will prototype ideas, integrate them into large-scale training systems, and contribute to publications.

The role requires hands-on experience with diffusion/autoregressive models, multi-view transformers, 3D representations, and strong coding skills. Work authorization in Switzerland is required.

Qualifications

• Candidate must be currently enrolled in or pursuing a PhD in computer vision, graphics, ML or related field.
• Strong hands-on experience with state-of-the-art model architectures for image/video reconstruction (e.g., diffusion/autoregressive, multi-view transformers).
• Proficiency with PyTorch and Python; solid understanding of 3D geometry and mathematical modeling.
• Proven track record in research with publications, grants, or patents or equivalent demonstrated impact.
• Excellent communication skills and ability to collaborate across groups and cultures.
• Work authorization in the country of employment for the duration of the internship/co-op.

Responsibilities

• Drive and participate in cutting-edge research for 3D/4D scene reconstruction, understanding and generation.
• Develop efficient deep neural network model architectures.
• Contribute research applicable to Meta wearables and AI development.
• Collaborate toward top-tier scientific publications with the team.

Skills

PyTorch

Python

Generative models

3D/4D reconstruction

Research internship

Education

PhD in computer vision / graphics / ML

Tools

PyTorch

Python

GitHub

CUDA

Job description

About

The Reality AI Research team explores, develops, and ships novel and cutting-edge technologies for 3D/4D scene reconstruction, understanding and generation, serving as the foundation for future wearables and AI products. Our team is tackling a variety of technical challenges across the entire tech stack, including but not limited to model architecture development, data curation, and large-scale model training. We're looking for candidates who share a passion for exploring and solving complex, challenging problems.We are looking for an exceptional PhD student with a proven track record in developing generative approaches for solving computer vision and graphics problems (e.g., conditional, generative models for videos or (dynamic) 3D content), as well as an outstanding software engineer who can efficiently prototype new ideas as well as integrate them into our large-scale data and training infrastructure.

Responsibilities

• Actively driving and participating in cutting edge research for 3D/4D scene reconstruction, understanding & generation.
• Developing efficient deep neural network model architectures.
• Contributing research that can be applied to Meta wearables and AI development.
• Together with our team, work towards a top-tier, scientific publication.

Minimum Qualifications

• Currently has, or is in the process of obtaining a PhD in the field of computer vision, computer graphics, machine learning, or a related field
• Hands-on knowledge with state-of-the-art model architectures (multi-view transformers, diffusion/autoregressive models, etc.,for image, object, scene, or video reconstruction, understanding, generation, etc.)
• Experience with PyTorch and Python
• Understanding of 3D geometry, statistical analysis of data, and mathematical modeling
• Excellent interpersonal and communication skills, cross-group and cross-culture collaboration
• High levels of creativity and quick problem solving capabilities
• Must obtain work authorization in the country of employment at the time of hire, and maintain ongoing work authorization during employment Intent to return to degree program after the completion of the internship/co-op
• Proven track record of achieving significant results as demonstrated by grants, fellowships, patents, as well as first-authored publications at leading conferences (e.g., SIGGRAPH, CVPR, ECCV, ICCV, NeurIPS, ICLR, ICML) or journals (e.g., PAMI, IJCV, JMLR, ToG)
• Experience with generative models (e.g., diffusion/autoregressive models), multi-view transformers (e.g., VGGT, MapAnything), 3D/4D representations (point clouds, Gaussian Splatting), video generation (e.g. WAN/Cosmos), and inverse rendering techniques
• Experience in 3D computer vision, graphics, and deep learning models for this domain
• Demonstrated software engineer experience via previous internships, work experiences, coding competitions, or widely used contributions in open source repositories (e.g. GitHub)
• Familiarity with AI native coding practices workflows