Machine Learning Engineer - Global E-commerce (ETA, Pricing & Conversion)

The Global E-commerce Algorithm team is the core engine driving the ultimate shopping experience and user growth for our rapidly expanding international platform. At the intersection of entertainment and commerce, we are shaping the future of "Discovery E-Commerce" by building cutting-edge AI, search, and recommendation systems that connect millions of users with the products they love. In this role, you will focus on solving massive-scale, consumer-facing challenges by bridging the gap between physical fulfillment and the digital user experience. You will build core algorithmic frameworks from the ground up, deeply integrating physical logistics factors—such as delivery speed, reliability, and shipping costs—into our core Search and Recommendation ecosystems. By dynamically optimizing impression allocation and conversion strategies, your work will directly drive user growth and platform GMV. You will push the boundaries of applied machine learning by leveraging cutting-edge techniques, including our proprietary OneTrans Model for ETA prediction, Causal Inference for dynamic pricing, and the latest advancements in Large Language Models (LLMs) and AI Agents. Responsibilities - Next-Gen ETA Prediction: Build and optimize end-to-end Estimated Time of Arrival (ETA) prediction systems leveraging our proprietary OneTrans Model. Analyze spatio-temporal sequences using deep learning to improve ETA accuracy, enhance consumer trust, and boost click-to-order rates. - Intelligent Pricing via Causal Inference: Design and implement intelligent shipping-fee pricing, free-shipping strategies, and subsidy-pricing coordination. Utilize advanced Causal Inference techniques to build multi-objective optimization models that perfectly balance user landed price, platform costs, and profitability. - Conversion & Impression Allocation: Deeply integrate key fulfillment factors (e.g., delivery speed, shipping cost) into core Search and Recommendation ranking algorithms. Design intelligent impression allocation and dynamic consumer presentation strategies that balance user experience and business costs, ultimately driving higher retention and repeat purchase rates. - LLM & AI Agent Innovation: Pioneer the development of domain-specific LLMs by leveraging massive e-commerce data for Continual Pre-Training (CPT), Supervised Fine-Tuning (SFT), and Reinforcement Learning (RL). Design and deploy intelligent AI Agents based on an "Agent + Skill" framework to autonomously diagnose and resolve complex user-facing and operational issues. Minimum Qualifications - Master's or PhD degree in Computer Science, Statistics, Mathematics, Operations Research, or a related highly quantitative field. - Solid foundation in data structures and algorithms, with proficient programming skills in Python, C++, or Java. - Strong fundamentals in applied machine learning, with hands-on experience processing large-scale data using Big Data tools (e.g., Spark, Hive SQL) and Deep Learning frameworks (e.g., TensorFlow, PyTorch). - Experience with A/B testing, experimentation design, and data-driven decision making. - Demonstrated ability to translate complex, open-ended business problems into highly scalable algorithmic solutions. - A strong passion for solving complex, high-impact e-commerce challenges and driving user growth. Preferred Qualifications - Deep expertise and industry experience in at least one of the following core areas: - Spatio-Temporal Modeling: Sequence prediction, ETA modeling, or graph neural networks. - Causal Inference: Dynamic pricing, uplift modeling, or user growth algorithms. - Search & Recommendation: RecSys ranking, computational advertising, or impression allocation. - LLM / Generative AI: LLM training pipelines (CPT, SFT, RLHF/RLAIF) and building Agentic workflows. - A strong track record of algorithmic innovation, demonstrated by publications at top-tier conferences (e.g., KDD, NeurIPS, WWW, SIGIR, WSDM, ICLR, ICML).