Job description
Responsibilities
Build and iterate a centralized exchange (CEX) full-chain risk control algorithm system based on artificial intelligence / machine learning technology, covering core scenarios such as account risk, trading risk control, anti-cheating, anti-money laundering (AML), fund security, abnormal behavior identification, and black industry confrontation.
Responsible for risk control sample mining, feature engineering, model training, tuning, and online deployment, using AI algorithms such as deep learning, traditional machine learning, graph algorithms, and time series models to construct user behavior profiles, risk scoring models, and anomaly detection models.
Design AI identification strategies and confrontation algorithms targeting black industry behaviors such as wash trading, airdrop farming, account theft, money laundering, malicious trading in contracts, batch account operations, and cross-account collusion, continuously improving the accuracy of risk control interception and reducing false positive rates.
Interface with business, operations, security, and compliance teams to break down risk control requirements, analyze risk cases and log data, and output algorithm solutions, strategy rules, and data conclusions.
Responsible for the full lifecycle management of risk control AI models: data preprocessing, model training, A/B testing, effect monitoring, online degradation warning, and regular iteration optimization to ensure the long-term stability and effectiveness of the models.
Research cutting-edge AI risk control technologies and new attack methods in the black industry, and implement innovative algorithm solutions based on the characteristics of CEX business to enhance the overall intelligence and automation level of risk control.
Collaborate with the engineering team to complete algorithm service packaging, interface debugging, and production environment deployment, ensuring high concurrency and low latency operation of the AI risk control system.
Requirements
Bachelor's degree or above in computer science, statistics, mathematics, artificial intelligence, data science, or related fields, with 2 years or more of practical experience in risk control algorithms / machine learning algorithms. Experience in finance, payment, internet black industry confrontation, and trading platform risk control is preferred.
Solid theoretical foundation in machine learning and deep learning, proficient in mainstream AI algorithms such as classification, clustering, anomaly detection, time series analysis, and graph learning, familiar with models such as logistic regression, XGBoost, LightGBM, CNN, RNN, Transformer, and Graph Neural Networks (GNN).
Proficient in Python and mainstream AI / data stacks: Scikit-learn, Pandas, NumPy, PyTorch/TensorFlow; capable of independently completing feature mining, model training, and experimental validation.
Experience in processing massive logs and behavioral data, understanding common risk control issues and optimization solutions such as feature engineering, sample selection, imbalance of positive and negative samples, and model generalization.
Familiar with the basic risk control system, knowledgeable in any field of anti-cheating, anti-money laundering, account security, and trading risk control, with experience in risk control algorithms for exchanges, financial platforms, and payment platforms being a plus.