Jan 2020 - Present
‧ Proposed data-driven business solutions to perform Anomaly/Fraud detection application in FinTech project.
‧ Collaborated with top financial company to define Business Understanding and establish the success criteria ; then proposed Data Science project lifecycle based on customers’ pain points.
‧ Implemented Supervised and Unsupervised learning methods(Random Forest, XGBoots, GMM, IsolationForest) and
also built a deep learning-based anomaly detection model(Autoencoder) with TensorFlow, then use these models for
outlier detection with cloud-based infrastructure datasets to find abnormal behaviors in Client’s IT platforms.
‧ Constructed data pipelines(ETL) on Customer’s Financial datasets to prepare the needed data for building the ML model to solve Insurance Fraud Detection project.
‧ Developed ML pipeline, including Exploratory Data Analysis(EDA), Data Preparation(Data Transformations, Feature Engineering), Model training and Model evaluation(cross validation, ROC-AUC, etc.), will generate ML solutions to meet clients’ business requirements.
‧ Optimised ML solution with different features(feature selection, feature importance, feature engineering), algorithms, modeling techniques, and hyperparameter tuning with decent metrics to get the best performing ML model.
‧ Interpreted trained ML models with Explainable AI tools and reported models’ behavior to our customer ; then visualized the final result with all relevant data through a dashboard created by Data Visualization tools(Matplotlib, Seaborn, Tableau).
‧ Handled FinTech project with an imbalanced data set provided by our client, using different/various techniques (class weighting, proper/suitable evaluation metrics, oversampling) to improve the performance of the ML solutions.