Responsibilities: 1. Utilize Convolutional Neural Networks (CNN) and multi-head attention mechanisms to identify asset price patterns and their developmental stages. Enhance recognition accuracy through visualization techniques and improve model robustness via data augmentation. Correlate pattern features with future returns/excess returns to construct pattern-based factors. 2. Combine feature engineering with genetic algorithms to train XGBoost/LightGBM/neural networks for asset return prediction, efficiently identifying effective factors to accelerate factor development. 3. Apply unsupervised learning methods including clustering analysis, Hidden Markov Models (HMM), and change-point detection to identify market regimes. Correlate findings with historical events (recessions, crises, inflation, trade wars) to model regime persistence and transition probability matrices. 4. Analyze factor performance across different market conditions. Develop dynamic portfolios using Markov Switching Models or rein