Regression-based ML project forecasting daily bike rentals using weather, calendar, and temporal features. Implements data cleaning, feature engineering, and model tuning.
Predictive modeling on the classic Titanic dataset, focusing on feature extraction, imputation, and comparative analysis between classic ML algorithms (Logistic Regression, Random Forest).
View ProjectUtilizes Recurrent Neural Networks (RNNs) to generate creative TV script text based on show transcripts, demonstrating sequence modeling skills.
View ProjectApplies Generative Adversarial Networks to synthesize realistic human faces from noise, showcasing knowledge of adversarial model architectures and training tricks.
Developed a financial sentiment analysis system using RoBERTa, fine-tuned on 5,800+ labeled financial sentences. The model classifies communications into negative, neutral, or positive to support equity research and investment decisions.
View Project