Responsible for the portfolio dynamic weight adjustment project, built the backtest framework, determined the evaluation system based on expected return and Barra decomposition, developed the weight generation strategy based on the optimizer and multi-stage rule based module, analyzed the influence of time interval, weight penalty and other parameters, and developed the online real-time trigger task

Responsible for the intelligent cover image distribution project, conducted hourly batch processing of cover image parameter data by Thompson Sampling Bandit distribution algorithm, SQL maintenance of hour-level increment and full model parameters, improved CTR by 1.2% in small traffic experiment

This three month deep learning internship at Meridian Global Inc. focuses on the application in quant market. My responsibilities mainly include building deep learning models to predict stock market prices, then developing trading strategies based on prediction results, and finally validating its profit potential.