Find your next role

Description

The North America Stores GenAI Evaluation Media (GEM) team is seeking an Applied Scientist to help shape the future of visual shopping experiences. We're building CXs and foundational capabilities to understand, enhance, and generate real-time GenAI imagery, videos and CXs that inspire customers and drive purchase confidence, towards our vision to be the leader in visual media. Specifically, the charter will focus on visual agentic experiences, multi-modal personalization, and real-time image/video generation, looking ahead as customer shopping continues to inspirational assistant-driven experiences.

As an Applied Scientist on the team, you will drive the research and development of agentic AI capabilities that inform and guide the customer's shopping journey through visuals. This includes building core science primitives for multimodal understanding, visual content generation and editing, personalized virtual try-on, and automated quality assurance. You will develop the foundational capabilities that enable customers to express and discover styles through multimodal conversation and receive personalized, visual responses that bring their ideas to life.

Your scientific approach will emphasize accurate, real-time visual understanding and generation, contextual understanding, and scalable personalization, enabling agentic AI to actively collaborate with customers to achieve their style goals. You will bring together computer vision, natural language processing, generative AI, and human-centered design to create agentic shopping experiences that are as intuitive as talking to a human specialist with a deep domain knowledge base. Success requires establishing robust metrics, collaborating with cross-functional partners, validating asset effectiveness across diverse customer touch points, and staying at the forefront of rapid advances in AI technology.

The ideal candidate will have deep technical expertise in Computer Vision, Generative AI, or related fields with a strong ability to connect scientific work to customer and business outcomes. You will partner with scientists, engineers, and stakeholders across Amazon to deliver innovation and uphold a culture of scientific excellence and customer obsession. This role requires both rigorous research skills and practical engineering instincts, with a focus on delivering solutions that scale. This is a unique opportunity to shape the future of visual commerce through applied AI research, building the systems that will define how hundreds of millions of customers discover and evaluate products and styles through visual experiences.

Key job responsibilities
Innovation & Technical Execution
• Develop core science primitives for vision and language understanding, visual content generation and editing, virtual try-on, and automated quality assurance via state-of-the-art computer vision, machine learning, and generative AI
• Design and implement visual agentic systems, balancing visual quality, relevance, latency, and cost
• Define metrics and success criteria for your scientific initiatives, ensuring rigorous validation across customer touch points
• Own end-to-end delivery of research initiatives from problem formulation through experimentation to production deployment
• Stay current with latest advances in AI/ML and identify opportunities to apply them to your problem space
• Drive development and deployment of scalable agentic systems for visual content understanding and generation
• Maintain high scientific and engineering standards in your work
• Tackle complex technical problems while maintaining practical focus on customer value
• Contribute to the team's culture of scientific excellence through presentations and publications at internal and external science forums

Cross-functional Collaboration
• Partner with product and engineering teams to deliver customer-facing features
• Collaborate with scientists and engineers across multiple teams within Amazon to align on technical approaches
• Communicate research findings and technical trade-offs clearly to both technical and non-technical stakeholders