We’re a London based tech start-up revolutionizing 3D content creation with MARSO, our breakthrough material capture technology. Using simple flash photography, MARSO precisely measures material properties of real-world objects, generating the world's most accurate, shader-agnostic PBR materials. This technology works on simple smart-phone scans all the way up to high end light-stage setups. This measured approach has established us as the industry standard for material accuracy, trusted by leading film studios and AAA game developers who rely on our technology for their most demanding projects.

Our current flash-based capture process delivers unparalleled accuracy, but we're now pushing boundaries further. Powered by our extensive library of measured material data, we're developing next-generation capture techniques that will make high-quality material acquisition accessible without flash photography.

As a rapidly growing startup with deep connections in film and gaming, we're uniquely positioned at the intersection of technical innovation and creative production. Our technology is not just used but actively sought after by major studios. We're making photorealistic 3D asset creation accessible to everyone - from major studios to independent creators - while maintaining the exceptional quality that has earned us the trust of the world's most demanding content creators.

As our Machine Learning Applied Researcher, you'll pioneer new approaches to material property prediction by leveraging and advancing state-of-the-art image generation techniques. You'll work with our unique dataset to develop a sophisticated ML pipeline that combines pre-trained models (like CLIP, SAM, and StableDiffusion), fine-tuned architectures, and custom diffusion models. This role offers the rare opportunity to push the boundaries of what's possible with image generation and understanding, working at the intersection of diffusion models, computer vision, and materials science.

You'll be developing novel approaches that might include:

• Creating bespoke diffusion models trained on our proprietary dataset
• Adapting and fine-tuning existing foundation models for material understanding
• Designing innovative architectures that combine frozen pre-trained models with custom components
• Building intermediate models for specific tasks like material segmentation or property classification

• Design and implement custom diffusion models for material property prediction
• Research and integrate state-of-the-art image generation models and techniques
• Create efficient training pipelines that leverage transfer learning from existing models
• Guide our data capture team on gathering targeted training data for specific use cases
• Research and implement latest advances in diffusion models, vision transformers, and multi-modal learning
• Collaborate with engineering teams to deploy research into production systems

• Strong experience with modern ML frameworks (PyTorch/JAX/TensorFlow)
• Proven track record implementing and training diffusion models
• Deep understanding of foundation models and transfer learning techniques
• Experience fine-tuning and adapting pre-trained models
• Expertise in computer vision and image processing
• Understanding of 3D graphics, rendering, and material properties
• Production-quality Python development skills
• Experience with large-scale ML training and deployment

• Creative problem-solving ability and research mindset
• Deep understanding of current trends in generative AI and diffusion models
• Self-directed while collaborating effectively with teams
• Clear communication of complex technical concepts
• Focus on scalable, production-ready solutions

Working at M-XR offers unique opportunities for professional growth:

• Access to exclusive, high-quality datasets for research and experimentation
• Opportunity to advance state-of-the-art in image generation and diffusion models
• Direct impact on cutting-edge technology used by major studios
• Leadership potential as our ML team expands