Yiru Fang Email & Phone Number

Urbana, IL, US

Washington, United States

Minneapolis, Minnesota, United States

• Work on the discovery of the internal principle of data exchangability across different environments, shedding light on the extension of conformal interval (independent and individually distributed data) to varying.
• The research goal is to identify the limitations and potential extensions of conformal interval techniques within the context of invariant risk minimization, aiming to enhance the robustness and reliability of machine.

• Took part in a research project focusing on developing a novel deep learning algorithm to predict kidney disease after cardiac surgeries from time-series health data.
• Conducted comprehensive data preprocessing on raw electronic health records (EHR) data to ensure data quality and integrity. Employed techniques such as data cleaning, missing value imputation, and feature engineering.
• Implemented and fine-tuned RNN, CNN, LSTM and GRU using TensorFlow and Keras to effectively capture temporal dependencies within sequential medical data, achieving a optimal 0.053 MSE in predicting kidney disease.
• Conducted evaluations and benchmarking of our proposed models against the traditional machine learning methods. Collaborated the result with domain experts to ensure the clinical relevance and interpretability of the.

• Applied mmdetection in high precision and small sample deep learning to recognize part depreciation
• Combine different backbone, neck and head algorithm like RCNN, FCNN, Cascade and created new model to get a best one
• Help creating a new self-developed algorithm by changing voidage and step combination, the mAP for new model increases by 143% compared to the original one
• Deploy the model to NVIDIA Triton, which increased the interactivity and made the model visual and usable for industrial application

Shenzhen, Guangdong, China

• Handled millions of e-commerce data, used SQL to analyze the crawled e-commerce data and provided customers with corresponding business recommendations
• Built an automation platform using Python, to monitor the accuracy of crawlers and make alerts
• Analyzed FMCG commodity data mainly from Unilever, L'Oreal and established a standardized database
• Analyzed the underlying structure of Kuaishou and China Tiktok to provide reference for crawer operation

Master Of Science - Ms, Statistics

Bachelor'S Degree, B.S. In Statistics, B.A. In Mathematics

Bachelor'S Degree, Data Science