Inference pipelines: Implement modular inference pipelines using SOTA Latent Diffusion Models to facilitate complex image-to-image pipelines.
Advanced Inpainting & Editing: Develop robust algorithmic solutions for targeted inpainting, outpainting, and coherent background generation within existing scenes.
Semantic Processing: Integrate SOTA segmentation models to automatically identify and isolate specific image regions for targeted visual modification.
Complex Chaining: Utilize custom Python backends to chain non-linear workflows, combining detection, segmentation, and generation steps.
Geometric & Structural Conditioning: Implement and fine-tune structural conditioning preprocessors to ensure generated elements strictly respect the vanishing points, perspective, and geometry of the source input.
Style Adaptation: Train and maintain parameter-efficient fine-tuning (PEFT) adapters to steer models toward specific high-value aesthetic styles and lighting conditions without catastrophic forgetting.
Automated Quality Scoring: Develop AI-on-AI scoring systems using advanced computer vision metrics to detect hallucinations, artifacts, or perspective failures before human review.
Requirements
MSc/PhD in Computer Science, AI, Computer Vision, or equivalent practical experience.
03+ years in Deep Learning/Computer Vision, with at least 1 year of industry experience specifically on Generative AI / Diffusion Models.
Mastery of Python and PyTorch; proficiency with the Hugging Face ecosystem.
Deep technical understanding of the Generative Stack, including Latent Diffusion Models, structural conditioning techniques, and VAEs.
Excellent prioritization skills, analytical and problem-solving.
Strong communication and presentation skills.
Resourceful, ability to work independently while balancing being part of a team.
Ability to use Gsuite tools in their work.
A strong portfolio of personal projects demonstrating your AI/ML skills (e.g., GitHub repositories, publications, conference presentations) is highly desirable.
Bonus points: Experience working intimately with complex image data or raw image data formats.
Vision specialists with strong generative experience.
Experience working in domains where truthfulness and structure are critical (e.g., architectural visualization, industrial simulation).
Experience creating custom node architectures for modular visual interfaces or contributing to open-source generative AI projects.
Tech Stack
Node.js
Python
PyTorch
Benefits
Competitive salary with attractive package of benefits incl. lunch, well-being, bike benefit