The rapid evolution of cryptocurrencies has re- shaped global financial systems, attracting both investors and researchers toward the prediction of their highly volatile price patterns. Accurate forecasting of cryptocurrency prices is es- sential for informed investment and risk management decisions. This research focuses on the use of Machine Learning (ML) and Deep Learning (DL) techniques to predict cryptocurrency prices, specifically Bitcoin and Ethereum, using a Long Short-Term Memory (LSTM) neural network. The model captures temporal dependencies in time-series data through sequential learning and minimizes prediction error using adaptive optimization techniques. Historical Open, High, Low, Close, and Volume (OHLCV) data are preprocessed and normalized for efficient model training. Experimental results show that the proposed LSTM model achieves an accuracy of over 98% (R2 score) and demonstrates robustness under dynamic market conditions. This study emphasizes the capability of ML-driven models in financial forecasting and suggests pathways for enhancing real-time crypto analytics and automated trading systems.