Description

The CreativeX RAPID(Real-time Ad Personalization & Insights Development) team is seeking passionate and talented SDM to join us. CreativeX is on a mission to enable brands of all sizes and categories to create, serve, measure, and optimize creative content with ease. Our team is responsible for tailoring the visual experience of ads to each context in real-time, leveraging nouvelle technologies such as latent diffusion models, large language models (LLM), reinforced learning (RL), computer vision, and related methods.

Our mission is to provide engaging, dynamically optimized creatives with low latencies both on and off Amazon for self-service brand advertisers. We automate the customization of product creatives with real-time catalog data while providing advertisers with the flexibility to customize their creatives according to their preferences.

Key job responsibilities
Job Responsibilities: Senior Machine Learning Engineer - Asset Sourcing & Insights

Overview
As an Senior Machine Learning Engineer on the CreativeX RAPID asset sourcing team, you will be responsible for building and optimizing the infrastructure that powers image and layout sourcing, while delivering actionable reporting and insights to drive creative performance.

Core Responsibilities

Asset Sourcing & Pipeline Development
• Design, develop, and maintain scalable ML pipelines for automated asset sourcing, including image and layout generation for selection
• Own the end-to-end Latte pipeline implementation for asset lifecycle management across multiple stages
• Implement and optimize asset indexing logic, creative variant ID generation, and asset candidate pool management
• Develop algorithms to improve asset quality, diversity, and relevance for real-time ad personalization
• Collaborate with upstream asset generation teams to ensure seamless integration and data flow

Machine Learning & Optimization
• Apply advanced ML techniques including computer vision, latent diffusion models, and reinforcement learning to enhance asset pool performance
• Build models to predict asset performance metrics (CTR, conversion rates) and optimize asset allocation
• Develop multi-modal optimization strategies combining images, product titles, and headlines
• Implement A/B testing frameworks and experimentation methodologies to validate model improvements

Reporting & Insights
• Design and build comprehensive dashboards and reporting systems to track asset lifecycle stages, leakage analysis, impression share, selection rates, and CTR performance
• Develop metrics and KPIs to measure asset sourcing efficiency and creative effectiveness
• Provide data-driven insights to stakeholders on asset performance, optimization opportunities, and pipeline health
• Create automated alerting systems for pipeline anomalies and performance degradation

Technical Leadership & Collaboration
• Mentor junior engineers and contribute to technical design reviews
• Partner with AI-Gen, AdFormat, Realtime Serving and Reporting teams to ensure alignment on technical solutions
• Drive best practices for ML model development, deployment, and monitoring
• Contribute to the technical roadmap and architecture decisions for the asset sourcing platform

System Performance & Scalability
• Ensure reliable asset sourcing pipelines to generate creative variants for real-time ad serving both on and off Amazon
• Optimize pipeline performance to handle large-scale asset processing
• Implement monitoring and observability solutions to maintain system reliability
• Design solutions that scale with growing advertiser demand and catalog complexity

Key Deliverables
• Production-ready ML models and pipelines for asset sourcing and insights
• Comprehensive dashboards and reporting tools for asset performance tracking
• Technical documentation and knowledge sharing with cross-functional teams
• Measurable improvements in asset quality, selection efficiency, and creative performance metrics

About the team
About the team
In an ideal future, AI Gen models have evolved to a state where entire creatives are generated on the fly and precisely tailored to the immediate preferences of each shopper. We bring this vision to life by sourcing a collection of pre-generated assets and assembling them, in real time, into personalized creative experiences for every customer. Our mission is to improve advertiser outcomes by tailoring the visuals to each shopper and publisher. The AI space is rapidly evolving and combining optimization with Amazon’s 1P audience data will be what enables us to leapfrog the competition. Through continuous creative optimization, we expect to learn how users interact with content across a variety of context. Our upstream teams will provide us an endless variety of content, informed by our learnings, via AI generation, transformation, and deconstruction. Advertisers will receive easily digestible insights to improve their future creatives. We succeed when shoppers see creatives personalized to their intent, publishers display creatives that match their unique look and feel, and advertisers benefit from increased performance.