Research Scientist – Efficient and Controllable Generative Models

Huawei envisions a world where technology connects people, empowers industries, and unlocks human potential. Guided by its mission to enrich lives through communication and intelligent innovation, Huawei stands at the forefront of global digital transformation. As a leader in Information and Communications Technology (ICT), the company pioneers breakthroughs in artificial intelligence, cloud computing, and smart devices - building the intelligent foundation of a fully connected world.

Through its Carrier, Enterprise, and Consumer business groups, Huawei delivers resilient digital infrastructure, advanced cloud and AI platforms, and transformative devices that enable progress at every level. Supporting 45 of the world’s top 50 telecom operators and serving one-third of the global population across more than 170 countries, Huawei is shaping a future where connectivity becomes a powerful catalyst for opportunity and sustainable growth.

About the role

The Computer Vision and Machine Learning Lab at Huawei Research Center in Zurich is looking for a Research Scientist to push the frontier of generative modeling and bring it to hundreds of millions of devices. You will conduct cutting-edge research on generative models for images, video, and 3D/4D content, with two goals at the center of your work: making these models efficient enough to run on-device, and making them predictable and controllable enough to be trusted in a product.

Generative models in consumer products must not only be fast; they must be reliable. Distorted faces, hallucinated or garbled text, altered identities, inconsistent geometry, and temporal flicker are unacceptable in a flagship camera or editing feature. You will develop methods that keep generation faithful to the input and to user intent, while running under real memory, latency, and power constraints.

The research is primarily applied to camera and imaging, image and video processing and understanding, and graphics: generative enhancement and editing of photos and videos, 3D-aware capture and reconstruction, dynamic scene understanding, and real-time rendering and content creation. Your work will directly shape next-generation Huawei consumer products, including flagship smartphones, laptops, tablets, and autonomous vehicles.

This is a role for someone who wants to publish at top venues and see their research ship.

What you will do

• Conduct original research on efficient generative models, including diffusion and flow-based models, few-step and one-step generation, distillation, quantization, and architectures co-designed with mobile and automotive hardware.

• Develop methods for controllable and reliable generation: precise conditioning and guidance, preservation of faces, identity, text, and scene structure, artifact detection and suppression, and consistency across frames and views.

• Extend these methods to 3D/4D generation, including neural and Gaussian scene representations, dynamic scene reconstruction, and controllable content creation.

• Apply the results to camera and imaging pipelines, image and video enhancement, editing and understanding, and graphics and rendering use cases on consumer devices.

• Transfer research results into Huawei products in close collaboration with product, hardware, and software engineering teams.

• Publish at leading conferences and journals (e.g., CVPR, ICCV, ECCV, NeurIPS, ICLR, SIGGRAPH) and contribute to the research community.

• Mentor interns and junior researchers, and collaborate with academic partners in Switzerland and abroad.

What we are looking for

• PhD (or equivalent research experience) in computer vision, computer graphics, machine learning, or a related field.

• Strong publication record in generative modeling, image or video restoration and editing, neural rendering, 3D/4D vision, or efficient deep learning.

• Solid hands-on experience with modern deep learning frameworks (e.g., PyTorch) and the ability to write clean, efficient research code.

• Deep understanding of diffusion and flow-based models, including sampling, guidance, conditioning, and distillation techniques.

• Curiosity, independence, and the drive to turn research into real-world impact.

Nice to have

• Experience with controllability and faithfulness in generation, such as identity or text preservation, structure-guided editing, or artifact evaluation.

• Experience deploying models on mobile, or embedded platforms (e.g., NPUs, quantization-aware training, on-device runtimes).

• Familiarity with 3D representations (meshes, point clouds, implicit fields, Gaussian splatting) and video or dynamic scene modeling.

• Background in autonomous driving perception, simulation, or world models.

Our team

You will join an international research team in the heart of Zurich, one of Europe's strongest hubs for computer vision and machine learning. Our team members hold PhDs from leading institutions such as ETH Zurich and EPFL and have a track record of publications at top-tier venues. We combine the openness and rigor of an academic lab with the resources and reach of a global technology company.

What we offer

• The opportunity to define the state of the art in a fast-moving field and see your work deployed at massive scale.

• Freedom to publish and to engage with the academic community.

• A collaborative, international environment in Zurich, with close ties to ETH Zurich, EPFL, and other leading research institutions.

• Competitive compensation and benefits.