岗位职责

1、协助团队完成美图全域用户画像体系建设，参与日常数据清洗、特征处理、标签生产等流程优化，配合保障画像数据准确性、完整性与时效性，支撑影像、AIGC、个性化推荐业务落地。

2、协助迭代用户兴趣、用户意图、用户分层等基础画像模型，基于多源用户行为数据挖掘用户核心偏好，协助产出标准化用户标签与用户向量，优化用户建模效果。

3、跟进大模型、Agent前沿技术在用户理解场景的应用探索，协助落地大模型增强用户画像的轻量化方案，助力AI Agent个性化能力升级。

任职要求

1、计算机、人工智能、统计学、数据科学等相关专业，硕士及以上在读学生，具备扎实的机器学习、数据挖掘基础理论功底。

2、熟练掌握Python编程语言，熟悉PyTorch、LightGBM、XGBoost等至少一种机器学习/深度学习工具框架，了解特征工程、用户建模基础流程。

3、对大模型、Agent技术有基础认知，对用户画像、搜索推荐、AI影像业务有浓厚兴趣，愿意深耕算法落地场景。

4、具备良好的数据敏感度、逻辑思维能力，善于通过数据发现问题，拥有较强的学习能力、沟通能力与团队协作意识，能稳定实习3个月及以上。

优先条件

1、有搜索推荐、用户画像、数据挖掘相关项目实习或竞赛经验者优先；

2、熟悉大模型微调、提示词工程、智能Agent基础应用者优先。

What You'll Be Part Of:

You'll join the team building Meitu's user profiling and user understanding layer across imaging, AIGC, and recommendation scenarios. We turn multi-source behavioral signals into reliable labels, embeddings, and models, and we're exploring how LLMs and agents can push user understanding further. This is an internship where you'll work with real production data and see your work affect actual products.

How You'll Contribute:

· Support the end-to-end user profile pipeline, including data cleaning, feature processing, label production, and workflow optimization, helping keep profile data accurate, complete, and timely for imaging, AIGC, and recommendation use cases;

· Iterate on foundational user models for interest, intent, and segmentation, mining core preferences from multi-source behavioral data to deliver standardized user tags and user embeddings;

· Explore how large language models and agent-based approaches can improve user understanding, and help implement lightweight solutions that bring LLM-enhanced profiles into AI Agent personalization.

What We Hope You Have:

· Currently enrolled in a master's or PhD program in computer science, artificial intelligence, statistics, data science, or a related field, with solid fundamentals in machine learning and data mining;

· Proficient in Python, with hands-on experience in at least one framework such as PyTorch, LightGBM, or XGBoost, and a working understanding of feature engineering and user modeling pipelines;

· Basic understanding of LLMs and Agent technologies, with genuine interest in user profiling, search/recommendation, or AI imaging and a desire to solve applied algorithm problems;

· Strong data intuition and logical thinking: you can identify issues through data and iterate, communicate clearly with the team, and commit to a stable internship of at least 3 months.

Even Better If You:

· Have project, internship, or competition experience in search/recommendation, user profiling, or data mining;

· Are familiar with LLM fine-tuning, prompt engineering, or basic intelligent agent applications.