Role overview
- Design, implement, and optimize image and video generation pipelines using state-of-the-art models to produce high-quality visual content at scale.
- Build and maintain multi-model generative workflows using orchestration tools that chain together generation, inpainting, upscaling, style transfer, and conditioning steps into production-ready pipelines.
- Fine-tune and adapt image generation models using techniques such as LoRA, DreamBooth, ControlNet, IP-Adapter, and textual inversion to achieve brand-consistent, style-controlled outputs.
- Leverage multimodal and vision-language models for image understanding, visual analysis, automated tagging, and quality evaluation within generative workflows.
- Evaluate, prototype, and integrate emerging video generation models into creative and product workflows.
- Develop agentic AI pipelines that orchestrate multi-step visual content creation, from prompt generation and image synthesis to post-processing and delivery.
- Collaborate with cross-functional teams including Creative, Product Management, and Engineering to translate brand and business needs into scalable generative AI solutions.
- Lead the technical evaluation of new generative AI models, tools, and vendors as the landscape evolves, influencing decisions for URBN's visual AI technology stack.
- Guide data curation and preparation strategies for fine-tuning, including dataset construction, annotation workflows, and synthetic data generation.
- Analyze and benchmark model outputs for quality, consistency, and brand alignment, designing robust validation and feedback loops that combine quantitative metrics with qualitative human assessment.
- Partner with engineers to translate research prototypes into production-grade services and APIs, with attention to cost optimization and throughput at scale.
Basic qualifications
- 5+ years of industry experience in data science, machine learning, or AI engineering, with a strong foundation in ML fundamentals.
- 1+ year of hands-on experience working with image generation models in a professional or serious applied context, not casual experimentation.
- Strong proficiency in Python, with practical experience using ML frameworks such as PyTorch and Hugging Face multimodal models.
- Hands-on experience with image model fine-tuning and conditioning techniques
- Working knowledge of GenAI workflow orchestration tools for building multi-step generation pipelines.
- Experience with multimodal and vision-language models for image understanding, captioning, or visual analysis.
- Experience with cloud-based AI infrastructure for training, fine-tuning, and serving generative models.
- Proven ability to evaluate and rapidly adopt new generative AI models and tools as the field evolves.
- Strong visual sensibility, with an eye for image quality, composition, and brand consistency in generated outputs.
- Excellent communication and collaboration skills, with the ability to bridge technical and creative teams.
- Bachelor's or Master's degree in a quantitative field such as Computer Science, Statistics, Engineering, or Mathematics, or equivalent practical experience.
Preferred qualifications
- Experience with video generation models and understanding of the evolving video GenAI landscape.
- Hands-on experience with creative design tools such as Adobe Photoshop, Firefly, or Figma, especially AI-augmented creative features like generative fill and inpainting.
- Experience building agentic AI workflows to orchestrate multi-model pipelines.
- Familiarity with fashion, retail, or e-commerce applications of generative AI, such as virtual try-on, AI product photography, or on-model image generation.
- Background in computer vision fundamentals like segmentation, detection, and embeddings that complement generative work.
- Experience with prompt engineering at scale, developing systematic prompt libraries or structured prompting strategies for consistent visual output.
Tags & focus areas
Used for matching and alerts on DevFound Fulltime Ai Data Science Generative Ai