• Developed an end-to-end machine learning solution to predict drug usage based on personality traits and demographic data.
‍
• Aimed to identify potential users to support targeted prevention and awareness campaigns.

• Used the UCI Drug Consumption dataset, applying models like Support Vector Machines (SVM), XGBoost, and Neural Networks (TensorFlow).
‍
• Applied techniques like dimensionality reduction PCA for feature simplification and UMAP for data visualization to uncover hidden patterns.
‍
• Focused on data preprocessing, feature engineering, and model optimization to achieve high recall.

• Built a model that effectively predicted drug usage patterns with high accuracy and recall.
‍
• Demonstrated strong skills in model training, evaluation, and building complete machine learning pipelines.

• Developed an end-to-end machine learning solution to improve long-term wind data correction for wind farm optimization.

• The goal was to replace a time-consuming neural network with a more cost-efficient regression model to predict wind conditions over a 20-year span.

• Applied Random Forest and regression models to predict wind data based on both short-term mast data and long-term meso data.
‍
• Focused on feature engineering, aligning data frequencies between mast and meso datasets, and optimizing model parameters.
‍
• Collaborated using GitHub for version control, allowing effective team communication and code management.

• Delivered a regression-based solution that matched the performance of the neural network at a significantly reduced computational cost.
‍
• Demonstrated skills in machine learning, data preprocessing, and real-world application of regression models for energy data optimization.