As a Machine Learning engineer, your responsibility is to design and implement training pipelines for ML models. Your current task involves creating an end-to-end training pipeline for a TensorFlow model. This model will be trained on several terabytes of structured data. The pipeline needs to include data quality checks before the training phase to ensure the integrity and accuracy of the data. Additionally, it must incorporate model quality checks after the training phase but before deployment to guarantee that the model meets performance standards. To meet business objectives, you need to minimize development time and reduce the necessity for infrastructure maintenance. Given these requirements, what is the best approach to build and orchestrate your training pipeline?