You are a Machine Learning Engineer at an e-commerce company with three machine learning models: a recommendation engine, a real-time fraud detection system, and a demand forecasting model. The recommendation engine requires low-latency predictions and dynamic scaling during peak times. The fraud detection system must process suspicious transaction events in real-time with variable workload. The demand forecasting model processes historical sales data in nightly batch jobs. Which AWS deployment strategy is MOST SUITABLE for each model?