Adaptive Machine Learning Data Pipeline Architecture

Design a state-of-the-art machine learning data pipeline that can automatically handle data quality issues, feature engineering, and model retraining. Create a system with built-in monitoring for data drift, concept drift, and model performance degradation. Develop a modular architecture that supports multiple machine learning algorithms and can dynamically select the most appropriate model based on incoming data characteristics.