岗位职责
】 智能广告竞价 (Bidding) 引擎建设：负责 Google/Meta 等海外效果广告媒体的竞价策略算法设计与落地。综合运用机器学习、运筹优化、强化学习等技术，设计全局智能出价与预算调配引擎，实现在总预算、ROI等多维业务约束下的全局最优决策自动化。 收益预估与弹性预测：深入理解机票、酒店等产线的业务逻辑，融合内外部动态特征，优化细粒度的转化率及收益预测模型；结合因果推断技术，建立广告调控的边际收益与弹性预测模型，精确预测不同出价设定下的流量、成本与订单弹性，实现精准的量级与ROI调控。 探索与利用机制设计：设计低成本的智能拓流与测款策略，科学量化系统不确定性，在高转化产品变现与长尾潜力探索间找到最佳平衡，全面挖掘全球市场流量与库存产品潜力。 选品与投放全链路优化：优化选品模型，打通选品准入与广告竞价的策略闭环，拓展算法在营销场景下的应用边界。 策略评估与落地：主导设计科学的策略实验与因果评估框架；负责策略的端到端开发落地与效果监控。 业务沟通与协同：将复杂的算法逻辑与数据表现转化为清晰的业务语言，跨团队高效协同，共同驱动业务增长。 【
职位要求
】 计算机科学、统计学、运筹学、应用数学或相关专业本科及以上学历。 3年及以上相关领域工作经验，有广告竞价算法、商业化变现、运筹优化等领域核心模块主导经验者优先。 熟练掌握常用机器学习与深度学习模型，熟悉 TensorFlow/PyTorch 等主流框架，在运筹优化、因果推断、强化学习中至少一个领域有较深积累与落地经验。 熟练掌握 Python/Java/C++ 等至少一门编程语言；熟练使用SQL/Spark等工具进行数据处理；具备科学的实验设计与评估分析能力。 优秀的业务与数据敏锐度，能迅速将复杂的业务痛点抽象为清晰的算法问题。面对不完美的现实数据系统（如延迟、稀疏），能设计稳健的降级与容错算法方案。 具备极强的ownership，为策略的业务效果负责；具备优秀的跨团队协同沟通能力与快速学习能力。 Job Responsibilities Intelligent Advertising Bidding Engine Development: Lead the design and implementation of bidding strategy algorithms for overseas performance advertising media including Google and Meta. Apply machine learning, operations research optimization, and reinforcement learning to design global intelligent bidding and budget allocation engines, achieving automated globally optimal decision-making under multi-dimensional business constraints such as total budget and ROI. Revenue Forecasting and Elasticity Prediction: Develop deep understanding of business logic across flight, hotel, and other product lines; integrate internal and external dynamic features to optimize fine-grained conversion rate and revenue prediction models. Leverage causal inference techniques to build marginal revenue and elasticity prediction models for advertising regulation, accurately forecasting traffic, cost, and order elasticity under different bidding settings to enable precise scale and ROI control. Exploration-Exploitation Mechanism Design: Design low-cost intelligent traffic expansion and product testing strategies; scientifically quantify system uncertainty; achieve optimal balance between high-conversion product monetization and long-tail potential exploration; fully unlock global market traffic and inventory product potential. Product Selection and Full-Chain Advertising Optimization: Optimize product selection models; close the strategic loop between product selection qualification and advertising bidding; expand algorithm application boundaries in marketing scenarios. Strategy Evaluation and Implementation: Lead the design of scientific strategy experimentation and causal evaluation frameworks; own end-to-end strategy development, deployment, and performance monitoring. Business Communication and Collaboration: Translate complex algorithm logic and data performance into clear business language; collaborate efficiently across teams to jointly drive business growth. Qualifications Bachelor's degree or above in Computer Science, Statistics, Operations Research, Applied Mathematics, or related fields. 3+ years of relevant experience; priority given to candidates with core module leadership experience in advertising bidding algorithms, commercial monetization, or operations research optimization. Proficient in commonly used machine learning and deep learning models; familiar with mainstream frameworks such as TensorFlow/PyTorch; deep expertise and proven implementation experience in at least one of operations research optimization, causal inference, or reinforcement learning. Proficient in at least one programming language including Python/Java/C++; skilled in data processing using SQL/Spark and other tools; capable of scientific experimental design and evaluation analysis. Excellent business and data acumen; ability to rapidly abstract complex business pain points into clear algorithmic problems. Capable of designing robust degradation and fault-tolerant algorithm solutions when facing imperfect real-world data systems (e.g., latency, sparsity). Strong ownership mentality, accountable for strategy business outcomes; excellent cross-team collaboration and communication skills; rapid learning ability.
