The aim to decrease the maintenance cost of generators used in wind energy production machinery. This is achieved by building various classification models, accounting for class imbalance, and tuning on a user defined cost metric (function of true positives, false positives and false negatives predicted) & productionising the model using pipelines.

This is a custom library for data processing, visualization and machine learning tools.

Advanced CIFAR-10 image classification using ResNet-inspired CNN with residual blocks, achieving 92%+ accuracy through comprehensive regularization, data augmentation, and professional ML engineering practices.

Split train/val/test coco dataset and Adjust annotations & categories.

This project predicts used car prices using a feedforward neural network regression model implemented in PyTorch. Features include car age, mileage, and other attributes. The pipeline supports feature normalization, train/validation/test splitting, and visualization of training and validation loss curves.

The aim is to decrease maintenance cost of generators used in wind energy production machinery. This is achieved by building various classification models, accounting for class imbalance, tuning on a user defined cost metric (function of true positives, false positives and false negatives predicted) & productionizing model using pipelines

Tire condition classification using ResNet and transfer learning. This project applies deep learning to identify whether a tire is in good or bad condition based on image data.

Classifying Travel Mode choice in the Netherlands using KNN, XGBoost, RF and TabNet

This project predicts loan approval outcomes (Approved/Rejected) using a PyTorch neural network. It includes data preprocessing, train/validation/test split, model training with BCEWithLogitsLoss, and inference with probability-based classification.