Principal Scientist, Generation Data Architect (Image & Video & Audio)

Adobe • San Jose, CA

3d

About The Position

We’re looking for a Principal Scientist (P60) to shape the data strategy behind Adobe Firefly’s multimodal foundation models (image, video, audio). In this role, you’ll work across research and engineering to improve how large-scale visual data is sourced, curated, and used to train high-quality generative models. Your work will directly impact how Firefly understands concepts like style, motion, and real-world variation—helping deliver more realistic, controllable, and useful creative tools to millions of users. This is a senior individual contributor role with broad influence across teams and model generations.

Requirements

• 10+ years of experience in ML, data systems, or AI research, including work on large-scale or foundation models
• Experience shaping data strategies, representations, or training approaches for generative or multimodal systems
• Expertise in image and/or video and audio generation, with a strong understanding of how data affects model behavior
• Understanding of how data, model architecture, and training dynamics interact
• Ability to communicate complex ideas clearly and work effectively with technical and cross-functional partners

Nice To Haves

• Master’s degree or Ph.D. in Computer Science, Machine Learning, or a related field preferred

Responsibilities

• Develop a long-term approach to generation-focused data that improves image, video, and audio synthesis at scale
• Guide decisions around data quality, diversity, and composition for foundation model training
• Explore new methods to address gaps in current data approaches and improve model performance
• Work closely with model teams to align data design with model architecture and training behavior
• Build and refine data curricula across training stages (what data is used, when, and at what scale)
• Run and interpret experiments to understand how data impacts quality, motion, robustness, and efficiency
• Combine organic, synthetic, and model-generated data to improve learning outcomes
• Partner with research, engineering, and product teams to guide data-related decisions
• Share insights that shape roadmaps across modeling, infrastructure, and applied research
• Support and mentor other scientists, contributing to a culture of strong experimentation and learning
• Help translate research ideas into scalable systems used in production models

What This Job Offers

• Job Type: Full-time
• Career Level: Principal
• Education Level: No Education Listed