End-to-End Image Captioning with Vision Transformer and GPT-2: Training and Deployment on AWS SageMaker