Zezheng L.

Quantitative Analytics Specialist @Wells Fargo

MN, US

Decision Science and Artificial Intelligence Validation - Natual Language Processing• Complete model validation activities for Virtual Channel, Home Lending and Auto Loans etc. Business Lines using Machine Learning / Deep Learning techniques.• Build complaint identification and compliant categorization models using customers’ call transcripts / chat bot transcripts data.• Create pipelines to convert unstructured transcripts data into vectors; build and tune the model to fit the data using GridSearch and Cross-Validation; compare the performance among different models and pick the top model based on the Business needs.• Build benchmarking models to challenge the developers’ models using advanced Deep Learning algorithms (CNN, LSTM etc.) and Large Language Models (BERT, Wav2vec, Whisper etc.).• Conduct independent testing on models’ weakness in terms of conceptual soundness, data quality, performance evaluation and monitoring process and write model validation report to clearly document the findings of the model.

University of Minnesota-Twin Cities

Master of Science (M.S.)

University of St. Thomas

Master's degree

University of Minnesota-Twin Cities

Bachelor of Science (B.S.)