EC2 Model Deployment & Metrics Monitoring

I already have a trained machine-learning model sitting in a repo; now I need it running reliably on AWS. Your first task is to containerise the model (Docker) and spin it up on an AWS EC2 instance—the environment I’ve chosen for this project. Once it is live, expose a secure REST endpoint so my application team can POST data and receive predictions in real time. Low latency matters, so please size the instance with that in mind and configure autoscaling policies if you feel they will help. Model quality can drift, so the next priority is a solid monitoring stack. I want to see Accuracy, Latency and Error rate streaming into CloudWatch (or Grafana/Prometheus if you prefer) with clear dashboards and alarms that alert us when thresholds are breached. Feel free to add extra metrics you deem useful, but those three must be front and centre. Wrap up with concise documentation: a read-me that shows me how to redeploy, how to hit the REST endpoint, and where to view/adjust the dashboards and alarms. Deliverables: • Containerised model deployed on AWS EC2 • Public, authenticated REST endpoint for inference • Live dashboards and alerts tracking Accuracy, Latency, Error rate • Step-by-step documentation and any IaC scripts (Terraform, CloudFormation or similar) Acceptance criteria: I can send a sample JSON payload to the endpoint, receive a valid prediction under the expected latency, and watch the three key metrics update in real time on the dashboard. When you reply, share one similar deployment you’ve handled, outline how you usually instrument model metrics, and give me a ballpark timeline so we can lock in milestones quickly.