Staff Machine Learning Engineer – Ads Retrieval

Staff Machine Learning Engineer – Ads Retrieval

Requirements

• MS or PhD in Computer Science, Statistics, or related field with a strong foundation in machine learning and information retrieval, and expertise across a range of retrieval modeling techniques.
• 6+ years of industry experience architecting, building, and scaling large-scale production recommendation or search systems, with a focus on high-performance retrieval leveraging diverse modeling approaches.
• Deep expertise in recommendation systems, especially large-scale retrieval algorithms and architectures, encompassing Generative Retrieval, User Sequence Modeling, Learning-to-Rank, and efficient ANN techniques.
• Mastery of deep learning techniques and proven ability to optimize model performance for complex retrieval tasks in large-scale environments, across various model types including generative, sequence-based, and ranking models.
• Demonstrated ability to lead complex technical projects across multiple areas of retrieval innovation, drive balanced technological advancements, and mentor junior engineers in a fast-paced, collaborative environment.
• Excellent communication and cross-functional collaboration skills, capable of articulating complex technical visions and building consensus across diverse teams, representing a comprehensive understanding of various retrieval technologies.

Responsibilities

• Design and implement a diverse portfolio of next-generation retrieval models for Shopping Ads: Pioneer advanced architectures beyond traditional approaches, becoming a leader in implementing and optimizing Generative Retrieval, User Sequence Modeling, and Learning-to-Rank models to significantly enhance ad relevance, capture user intent, and improve ranking quality.
• Build and optimize massively scalable and efficient Ads Retrieval infrastructure: Lead the evolution of our next-gen infrastructure, capable of handling a 5 billion+ Shopping Ads index, ensuring lightning-fast, cost-effective retrieval through techniques like efficient ANN algorithms, GPU-accelerated systems, and embedding quantization.
• Drive innovation in personalized Shopping Ads recommendations through advanced modeling: Develop hyper-personalized retrieval models that incorporate user sequence modeling to deeply understand user shopping journeys and leverage learning-to-rank to surface the most relevant ads, while also exploring the potential of generative retrieval for novel ad discovery.
• Champion a holistic approach to retrieval excellence: Evaluate and integrate a range of cutting-edge technologies, including Large Language Models (LLMs), Generative Retrieval techniques, advanced Sequence Models, and efficient ANN algorithms, to continuously revolutionize Shopping Ads retrieval and push the boundaries of relevance, efficiency, and user engagement.
• Collaborate cross-functionally to optimize the entire Pinterest Shopping Ads ecosystem: Partner with Product, Data Science, and Engineering teams to holistically improve the user journey, optimize ad performance across all stages of retrieval and ranking, and drive demand-side growth for Shopping Ads, ensuring a balanced approach across different modeling and infrastructure innovations.

Preferred Qualifications

• Hands-on experience developing and deploying recommendation systems utilizing Generative Retrieval, User Sequence Modeling, and/or Learning-to-Rank techniques.
• Expertise in computational advertising, particularly within Shopping Ads or e-commerce domains, with a broad understanding of different retrieval modeling paradigms.
• Proven track record of optimizing GPU-based systems for high-throughput, low-latency retrieval and experience in implementing embedding quantization and other efficiency techniques.
• Familiarity with a wide range of retrieval efficiency and scaling techniques, including efficient ANN algorithms, token-based retrieval, and embedding quantization.