Machine Learning – Recommendation

Key Responsibilities

• Lead the technical design and continuous optimization of large-scale recommendation systems, covering retrieval, pre-ranking, ranking, re-ranking, and blending.
• Diagnose effectiveness and efficiency bottlenecks across the recommendation pipeline, improving end-to-end performance through training data, features, model architecture, optimization objectives, and serving architecture.
• Design optimization objectives focused on user engagement and long-term value, including DAU, retention, session duration, and visit frequency, rather than focusing solely on short-term clicks, conversion, or GMV.
• Design training labels, loss functions, negative sampling and sampling strategies, multi-objective optimization methods, and evaluation frameworks, and validate their impact through online experiments.
• Optimize large-scale model training and online inference, balancing model quality, latency, throughput, infrastructure cost, and system reliability.
• Contribute to the development and productionization of Semantic ID, generative recommendation, and generative sequence modeling, continuously improving model quality, engineering efficiency, and business impact.
• Partner with product, engineering, and data teams to translate business problems into machine learning solutions that are testable, scalable, and production-ready.
• Stay current with advances in recommendation systems, assess their practical value, and drive promising technologies into production.

Requirements

• Hands-on experience in search ranking, recommendation systems, feed algorithms, homepage optimization, or related areas.
• Deep understanding of the end-to-end recommendation pipeline, with experience optimizing across retrieval, ranking, re-ranking, and online serving.
• Strong machine learning fundamentals, with the ability to design loss functions, training labels, feature systems, sampling strategies, multi-task or multi-objective learning approaches, and model evaluation methodologies.
• Experience with large-scale production machine learning systems, including using data and experiments to diagnose problems and taking solutions from prototype to production.
• Strong understanding of user engagement, retention, and long-term value optimization, including the trade-offs between short-term business metrics and long-term user value.
• Strong technical judgment and cross-functional collaboration skills, with the ability to define technical direction and drive complex projects to completion.
• Good command of written and spoken English.

Nice to Have

• Research or hands-on experience with Semantic ID, recommendation tokenizers, vector quantization, Residual Quantization, or hierarchical item encoding.
• Research or hands-on experience with generative retrieval, generative ranking, or unified generative recommendation architectures.
• Familiarity with Transformer-based user behavior sequence modeling, autoregressive generation, or constrained decoding.
• Experience modeling extremely large item or content catalogs, or long user behavior sequences.
• Experience with cold-start recommendation, new-user recommendation, or cross-domain recommendation.
• High-quality publications in recommendation systems or generative recommendation, or experience leading production projects with significant business impact.