Predicted pCO₂ and fCO₂ using XGBoost (DMatrix API) on SOCAT datasets, analyzing CO₂ dynamics at various missing data rates ranging from 15% to 90% for the Arabian Sea & Bay of Bengal.
Built a comprehensive ML pipeline utilizing Random Forest (achieving 90% accuracy) for churn prediction using client and price datasets.
Technology used: NumPy, Pandas, Matplotlib, Seaborn, Scikit-learn, and Random Forest.
Port Operations Automation | CNN + MobileNetV2
Deep Learning | IIT Kanpur
01.2025 - 02.2025
Developed a CNN to classify nine types of boats, achieving 76.29% accuracy.
Optimized with MobileNetV2 for mobile deployment, improving accuracy to 97.12% with early stopping over 50 epochs.
Technologies used: NumPy, Matplotlib, TensorFlow, Keras, Max Pooling, Global Average Pooling, Early Stopping, Dropout, Batch Normalization, MobileNetV2.
Employee Turnover Analytics
Machine Learning | IIT Kanpur
11.2024 - 12.2024
Built three ML models using Logistic Regression, Gradient Boosting Machine, and Random Forest (achieved 99.2% accuracy) to classify left (or stay) employees using SMOTE-balanced HR data, and derived retention strategies from clustering.
MEDICAL INSURANCE PREMIUM PREDICTION ML MODEL at Python(NumPy, Pandas, Scikit-learn), Machine LearningMEDICAL INSURANCE PREMIUM PREDICTION ML MODEL at Python(NumPy, Pandas, Scikit-learn), Machine Learning
Casual Academic Demonstrator for EESC101/DSCI105 Planet Earth, EESC202 Shaping Earth’s Surface, EESC323 Advanced River Processes at University of WollongongCasual Academic Demonstrator for EESC101/DSCI105 Planet Earth, EESC202 Shaping Earth’s Surface, EESC323 Advanced River Processes at University of Wollongong