Yudong Chen Email and Phone Number

Develop and validate commercial loan risk rating scorecards.Manage back-testing and ongoing monitors for all commercial loan risk rating scorecards covering entire commercial loan portfolio (700+ Billion USD).Manage the calibration and back-testing for wholesale banking lost forecast models as part of CCAR submission.Working with vendors such as Moodys and BlackRock on RiscCalc, Commercial Real Estate Loss Forecast models.Building coding/data infrastructure for commercial loan risk rating scorecards and wholesale loss forecast models.

In charge of the design and maintenance of Standard Modeling Tools and Standard Variable Pull Tool and Text Analytics tool. These are the standard modeling tools used widely used across the modeling and analytics departments in Bank of America. (2 patents approved and 3 pending approved associated with the various tools designed)Take primary responsibility in the model management for the 300+ models used by different LOB in consumer banking, including mortgage, heloc, credit card, deposit, auto, etc. Such activities include model performance tracking, change control, data migration, input and output stability, etc. In charge of the design, development and execution of automated model validation and model ongoing monitoring, working closely with model risk management and modelers across different lines of business.In charge of the training of new modelers and supporting of existing modelers with regard to technology and methodology needs.Conduct advanced R&D projects leveraging most advanced statistical method and technology, including implementation of Random Forest of Boosting algorithms, SAS high performance analytics testing, hadoop testing, etc.

Build models for various risk management and marketing purposes, such as cash strategy, credit line increase, and customer segmentation. Responsibilities include model development, legal and technical review, implementation and validation. Techniques used include logistic regression, linear regression, optimization and decision tree based segmentation, etc.. Design and implement testing/control block for analysis and modeling purpose. Responsibilities include designing various testing blocks and tracking the (longitudinal) performance and eventually building model and conducting analysis based on the performance of testing blocks vs. control block.In charge of the design and maintenance of Standard Modeling Tools and Standard Variable Pull Tool which is widely used across the modeling and analytics departments in Bank of America. Take primary responsibility in the model management, including automated model validation, tracking and summarization.In charge of managing and tracking credit bureau migration and UDAP (Card ACT) related impact over all models (200+) in Bank of America.

Modeled credit risks for different market segments to assist in strategy making (with SAS and SQL). Monitored the performance of Dscore (the major risk score of the company), analyzed the performance of Dscore and conducted the redesign of Dscore. Assisted in the strategy making of Credit Line Increase and Credit Line Decrease. Analyzed the impact of sub-prime mortgage crisis over the current portfolio and designed strategies to minimize the risk (high risk mortgage queue). Developed and maintained the MIS for the major types of risk trends of the current portfolio (With SAS and VB); developed new tools to facilitate the strategy making and modeling process (functioning and GUI implemented in Python). Conducted geographic analysis at the level of state, MSA and county, with the focus on both internal risk measures and macro economic measures (data source Moody).

Worked as the major responsible research associate for the Early Trend Detection Project.Took primary responsibility for the automatic processing of large scale data from nation-wide agent reports. Applied text-processing, text-mining and statistical modeling (temporal-spatial model) techniques in order to make short-term and long-term prediction and control of associated risks. Python and C++ were implemented.

Responsible for the experiment design and data analysis for psycholinguistic studies with Professor Mack in at Urbana Champaign. Major duties include experiment design for cross-sectional and longitudinal studies, data collection, and statistical analysis and programming on her research projects. SAS and SPSS were used to apply various statistical tests.