- Explore behaviors and distributions of common customers across OneMount and Techcombank. - Develop strategies to engage the remaining customers with similar behaviors. - Conduct ad hoc analysis on demand for OneMount customers. - Process data from over 10 million OneMount customers. - Model CASA account acquisitions through data analysis, probability modeling, lead generation, A/B testing, and production-ready model implementation. - Achieve a 1000% increase in CASA acquisitions compared to the benchmark. - Model and cross-sell credit card products to acquire new OneMount customers.

Lead 3 members: 2 data scientist and 1 data analyst Understand business problems and define the evaluation metrics Adhoc and experimental data analysis and A/B testing Identify a viable business model from an economic perspective to serve customers Label household income, feature engineering, probabilistic explainable modeling, evaluating benchmark and machine learning model Build structure of the project, divide tasks following scrum process Cross-check and merge code from members Improve approximately 20% on specific evaluation metrics Achieve approximately 60% Total Operating Income (TOI)

- Design a predictive classification model to differentiate between A-book investors, who pose higher risks to brokers, and B-book investors, who pose lower risks. - Transfer positions from A-book investors to the market makers and manage B-book investors' risk for profit optimization using models such as Neural Network, XGBoost, CatBoost, and Random Forest. - Explore retail investor behavior patterns using machine learning and technical indicators like moving averages, Bollinger Bands, and MACD. - Simulate profit and loss from classified investors and improve performance by 60% compared to benchmarks. - Enhance profit loss simulations by analyzing the sensitivity of bid-ask spread term structures, cover threshold term structures, and strategies for taking profit and cutting loss. - Analyze aspects such as return, volatility, bid-ask spread, and number of ticks to understand trend causality. - Detect traders using prohibited programming methods via models like Auto-encoder, KNN, and Isolated Forest. - Employ genetic algorithms to handle a large volume of investors.

Lead and manage a team of 2 machine learning engineers. Conduct data collection and exploratory data analysis (EDA). Perform feature engineering and model development. Implement evaluation metrics and monitoring. Conduct A/B testing for validation and performance measurement. Develop a multi-model endpoint on SageMaker for store-specific customization. Design and implement ETL data flow and feature storage using Airflow. Create RESTful services for real-time prediction employing FastAPI. Achieve improvement of 60–80% on evaluation metrics. Establish CI/CD pipeline using Bitbucket Pipeline, ECR, ECS, and SageMaker.

Building a website tailored for rental property needs. Developing a management system for rental operations. Creating a financial system to manage rental finances. Consulting on technologies and tools to optimize processes.

Applied Genetic Algorithm and Linear Programming to improve distance cost. Optimized the number of vehicles used in routing. Managed the distribution of weight per vehicle to enhance efficiency.

Collect, process, and mine data to analyze user behaviors. Personalize user experiences by analyzing and forecasting customer behavior through data-driven analysis. Construct a hybrid recommendation system utilizing popularity, content-based, and collaborative filtering using implicit and explicit data. Build a REST API for the recommendation system using the Django framework. Update periodic tasks for retraining the model.

Reduce drying time by controlling drum rotation, heating switch, and fan velocity using Rainbow, which includes seven extension models of Deep Reinforcement Learning. Analyze dryer data, such as temperature and humidity, to provide domain knowledge essential for building the Rainbow model.

Build a recommendation system for retail products using Amazon SageMaker. Predict customer buying frequencies. Anticipate the level of customer loyalty.