高级数据科学家 - 反作弊
Senior Data Scientist - Anticheating
职位描述 我们正在构建下一代基于 AI 的反作弊系统，用于竞争性策略游戏。
与传统的欺诈检测不同，我们的挑战位于以下领域的交叉点：
- 🎮 游戏 AI 与玩家行为建模
- 🧠 强化学习与决策系统
- 🔍 在对抗条件下的异常检测
你将负责在复杂环境中识别不明显的战略作弊行为，这些环境中的玩家会主动适应检测系统。这不是基于规则的检测——这是大规模的行为智能。
你将负责的工作 1️⃣ 行为建模与检测
- 设计机器学习/深度学习模型来检测作弊模式
- 建模玩家行为序列、策略和异常
- 构建系统来区分：
- 高技能操作 vs. AI 辅助操作
- 自然波动 vs. 利用行为
2️⃣ 反作弊系统设计
- 开发可扩展的检测流程（离线 + 实时）
- 从游戏日志/事件流中构建特征系统
- 设计检测准确性和鲁棒性的评估框架
3️⃣ ML / DL / 高级技术
- 应用并实验以下技术：
- 序列建模（RNN / 基于 Transformer）
- 异常检测
- 基于图或行为嵌入
- 探索与其他领域的交集：
- 强化学习
- 博弈论建模
- 对抗性机器学习
4️⃣ 与 AI 和工程团队合作
- 与以下团队紧密合作：
- 游戏 AI / RL 研究人员
- 后端/数据工程团队
- 将模型转化为生产系统
我们寻找的人选
✅ 核心要求
- 4年以上数据科学/机器学习相关经验
- 扎实的基础包括：
- 深度学习
- 统计建模
- 有以下任一经验：
- 欺诈检测/AML
- 风险建模
- 异常检测
- 行为分析
✅ 强烈信号（加分项）
- 有以下经验：
- 序列模型（LSTM / Transformer）
- 大规模行为数据
- 实时检测系统
- 有以下接触：
- 强化学习
- 游戏 AI
- 对抗系统
✅ 技术栈
- Python（必须）
- PyTorch / TensorFlow
- SQL / 数据流水线
- 有处理大规模数据集的经验
为什么这个职位有趣
- 🚀 解决类似大规模欺诈检测的问题——但更具挑战性
- 🎯 直接影响真实金钱/竞技环境
- 🧠 结合：
- 机器学习研究
- 生产系统
- 游戏 AI
- 🌍 全程远程，全球分布式团队
地点与签证
- 🌏 以远程为主（全球团队）
查看英文原文
About the Role We are building a next-generation AI-driven anti-cheating system for competitive strategy games.
Unlike traditional fraud detection, our challenge sits at the intersection of:
- 🎮 Game AI & player behavior modeling
- 🧠 Reinforcement learning & decision systems
- 🔍 Anomaly detection under adversarial conditions
You will work on identifying non-obvious, strategic cheating behaviors in complex environments where players actively adapt to detection systems. This is not rule-based detection — this is behavioral intelligence at scale.
What You’ll Do 1️⃣ Behavioral Modeling & Detection
- Design machine learning / deep learning models to detect cheating patterns
- Model player behavior sequences, strategies, and anomalies
- Build systems that distinguish:
- high-skill play vs. AI-assisted play
- natural variance vs. exploitation
2️⃣ Anti-Cheating System Design
- Develop scalable detection pipelines (offline + real-time)
- Build feature systems from gameplay logs / event streams
- Design evaluation frameworks for detection accuracy & robustness
3️⃣ ML / DL / Advanced Techniques
- Apply and experiment with:
- sequence modeling (RNN / Transformer-based)
- anomaly detection
- graph-based or behavioral embeddings
- Explore intersections with:
- reinforcement learning
- game-theoretic modeling
- adversarial ML
4️⃣ Collaboration with AI & Engineering Teams
- Work closely with:
- Gameplay AI / RL researchers
- Backend / data engineering teams
- Translate models into production systems
What We’re Looking For
✅ Core Requirements
- 4+ years in Data Science / Machine Learning roles
- Strong foundation in:
- deep learning
- statistical modeling
- Experience in one or more of:
- fraud detection / AML
- risk modeling
- anomaly detection
- behavioral analytics
✅ Strong Signals (Big Plus)
- Experience with:
- sequence models (LSTM / Transformer)
- large-scale behavioral data
- real-time detection systems
- Exposure to:
- reinforcement learning
- game AI
- adversarial systems
✅ Technical Stack
- Python (must)
- PyTorch / TensorFlow
- SQL / data pipelines
- Experience working with large-scale datasets
Why This Role is Interesting
- 🚀 Work on problems similar to fraud detection at scale — but harder
- 🎯 Direct impact on real-money / competitive environments
- 🧠 Blend of:
- ML research
- production systems
- game AI
- 🌍 Fully remote, globally distributed team
Location & Visa
- 🌏 Remote-first (global team)
- 🇯🇵 Japan relocation supported (visa sponsorship available for qualified candidates)
Who This Role is Perfect For
- Data scientists bored with “dashboard ML”
- Fraud / AML experts who want more complex, adversarial systems
- ML engineers who want to work closer to decision intelligence & behavior modeling
Originally posted on Himalayas