Run end-to-end machine learning pipeline

Execute end-to-end ML pipeline with stage verification

Reproducible end-to-end execution after configuration or code changes.

• · config file path
• · raw data directory
• · pipeline stages

• · processed data
• · engineered features
• · trained checkpoint
• · metrics JSON

• · pandas
• · PyTorch
• · scikit-learn
• · DVC (optional)

data/ directory exists, config.yaml valid

• · Early stage failure halts pipeline
• · missing dependencies
• · data corruption

• · Sequential error halt
• · stage output verification