Ni A. Email and Phone Number

• Built Natural Language Processing (NLP) and unsupervised Machine Learning (ML) algorithms, such as word embedding, K-Means clustering, to mine and quantify automation opportunities from conversation transcripts• Built deep learning based image classification models to identify facilities in technician submitted images of service installation• Built LSTM-based Deep Learning (DL) model for technician notes classification to predict causes of technical failures• Built NLP models to predict user reviews (i.e., sentiment) based on conversation transcripts• Applied decision-tree models, such as LightGBM, random forest, to detect frauds or mobility churns

Research assistant in the ECE department at Drexel.

Built deep learning anomaly detector (such as deep autoencoder, stacked denoising autoencoder), and classic anomaly detection methods (such as multivariate correlation analysis and robust principal component analysis (PCA)) to detect anomalies and attacks in network flow data

Conducted network security tests such as vulnerability scanningImproved and revised system-wide dashboard to display compliance trend and security risk level