TensorFlow Data Pipeline Builder

Build an optimized TensorFlow data pipeline for my machine learning project.

Data details:
- Data source: [FILES/DATABASE/API/TFRecords]
- Data size: [APPROXIMATE SIZE]
- Data types: [IMAGES/TEXT/TABULAR/TIME SERIES]
- Batch size needed: [BATCH SIZE]

Requirements:
1. Efficient data loading and parsing
2. Data augmentation transformations
3. Prefetching and caching strategies
4. Parallel processing configuration
5. Train/validation/test splitting
6. Handling imbalanced datasets
7. Memory-efficient processing for large datasets
8. Integration with model.fit()

Include performance benchmarking code to compare different pipeline configurations.