• Prototyped bond default prediction model based on mutual funds quarterly holding position information• Gained accuracy 93.1%, type I error 6.88%, type II error 10.53%, average alert time 627 days before default events• Establishing Bond Default Early Warning System by collecting finance, price, investors' behavioral, sentiment data and implementing machine learning algorithm• Improving financial fraud identification model's performance

• Projected future capital shortfalls for an urban commercial bank based on Basel III, determined the amount of IPO funds raised and issue price by comparison with peer companies’ relative valuation in A-share market and completed project proposal• Designed backdoor listing scheme for a state-owned enterprise with asset replacement and private placement and calculated value of injected assets, additional shares issued, ratio of shares held and market value according to target firm’s financial situation• Evaluated a financial group’s proposal for share swap between its own bank and some H-share listed insurance companies in terms of shares held, equity concentration as well as future prospects after transaction and formed financial consultant report

• Gathered daily financial events and summarized financial events analysis methods by reading investment banks’ research reports• Analyzed counterparts’ intelligent investment research products and wrote reports on functions, user experience and technologies• Trained RNN, LSTM model to predict daily close price of tech giant companies’ stocks like Apple Inc. (Tensorflow, Keras)