Crypto trading prediction Tool. Time series AI Project
Developed a machine learning pipeline to predict short-term crypto price movements using multivariate time series data. While not a labeling project in the strict sense, this initiative involved curating and preprocessing large datasets, creating labeled training targets based on price movement thresholds, and validating prediction outputs. Worked with python (PyTorch and pandas) to structure time windows, engineer relevant features, and format data for supervised learning. Also evaluated model predictions against real market outcomes and refined the training set accordingly. This project strengthened my understanding of dataset quality, label logic, and evaluation metrics all highly transferable to data labeling and AI training tasks.