- Google ML Bootcamp Kaggle competition: TOP 15 (from 314 participants) with 0.9337 accuracy. Kaggle notebook, where I've used the Tensorflow library to predict Yoga Pose by images using custom CNN and Transfer learning (EfficientNet, DenseNet) with fine-tuning.
- Loan Default Prediction Coursera Competition: Top 100 with 0.77 accuracy. I've used different algorithms and libraries:
- Scikit-learn for Random forest, KNN, and SVM algorithms with data transformation pipelines of preprocessing.
- Tensorflow for neural networks.
- Natural Language Processing with Disaster Tweets Kaggle competition - predicted which Tweets are about real disasters and which ones are not with 0.78 final score.
- Classic competitions:
- Digit recognizer - Tensorflow with CNN.
- Titanic with Neural Networks - Tensorflow library.
- Titanic with classic ML - Scikit-learn library usage Random forest, KNN, SVM.